Home/Federal/Federal Register/2026-08455

Notice2026-08455

ChemoCentryx, Inc.; Proposal To Withdraw Approval of New Drug Application for TAVNEOS (Avacopan) Capsule, 10 Milligrams; Opportunity for a Hearing

Primary source

Metadata and text below are from the Federal Register, a public-domain U.S. government work. Always verify the official published version before relying on it for any legal matter.

Published

April 30, 2026

Issuing agencies

Health and Human Services DepartmentFood and Drug Administration

Abstract

The Food and Drug Administration's (FDA, Agency, or we) Center for Drug Evaluation and Research (CDER) is proposing to withdraw approval of the new drug application (NDA) for TAVNEOS (avacopan) capsule, 10 milligrams (mg), held by ChemoCentryx, Inc., One Amgen Center Dr., Thousand Oaks, CA 91320 (ChemoCentryx or applicant), and is announcing an opportunity for the applicant to request a hearing on this proposal. The grounds for the proposal are twofold: (i) on the basis of new information before FDA, evaluated together with the evidence available to FDA when the application was approved, there is a lack of substantial evidence that the drug will have the effect it purports or is represented to have under the conditions of use prescribed, recommended, or suggested in its labeling; and (ii) the application contains untrue statements of material fact.

Full Text

<html>
<head>
<title>Federal Register, Volume 91 Issue 83 (Thursday, April 30, 2026)</title>
</head>
<body><pre>
[Federal Register Volume 91, Number 83 (Thursday, April 30, 2026)]
[Notices]
[Pages 23278-23288]
From the Federal Register Online via the Government Publishing Office [<a href="http://www.gpo.gov">www.gpo.gov</a>]
[FR Doc No: 2026-08455]

-----------------------------------------------------------------------

DEPARTMENT OF HEALTH AND HUMAN SERVICES

Food and Drug Administration

[Docket No. FDA-2026-N-1321]

ChemoCentryx, Inc.; Proposal To Withdraw Approval of New Drug
Application for TAVNEOS (Avacopan) Capsule, 10 Milligrams; Opportunity
for a Hearing

AGENCY: Food and Drug Administration, HHS.

ACTION: Notice.

-----------------------------------------------------------------------

SUMMARY: The Food and Drug Administration's (FDA, Agency, or we) Center
for Drug Evaluation and Research (CDER) is proposing to withdraw
approval of the new drug application (NDA) for TAVNEOS (avacopan)
capsule, 10 milligrams (mg), held by ChemoCentryx, Inc., One Amgen
Center Dr., Thousand Oaks, CA 91320 (ChemoCentryx or applicant), and is
announcing an opportunity for the applicant to request a hearing on
this proposal. The grounds for the proposal are twofold: (i) on the
basis of new information before FDA, evaluated together with the
evidence available to FDA when the application was approved, there is a
lack of substantial evidence that the drug will have the effect it
purports or is represented to have under the conditions of use
prescribed, recommended, or suggested in its labeling; and (ii) the
application contains untrue statements of material fact.

DATES: The applicant may submit a written request for a hearing by June
1, 2026 and submit all data, information, and analyses in support of
the hearing request by June 29, 2026. Others may submit electronic or
written comments by June 29, 2026.

ADDRESSES: The request for a hearing may be submitted by the applicant
by either of the following methods:

Electronic Submissions

Submit electronic comments in the following way:
<bullet> Federal eRulemaking Portal: <a href="https://www.regulations.gov">https://www.regulations.gov</a>.
Follow the instructions for submitting comments to submit your request
for a hearing. Comments submitted electronically to <a href="https://www.regulations.gov">https://www.regulations.gov</a>, including any attachments to the request for a
hearing, will be posted to the docket unchanged.

Written/Paper Submissions

Submit written/paper submissions as follows:
<bullet> Mail/Hand delivery/Courier (for written/paper
submissions): Dockets Management Staff (HFA-305), Food and Drug
Administration, 5630 Fishers Lane, Rm. 1061, Rockville, MD 20852.
<bullet> Because your request for a hearing will be made public,
you are solely responsible for ensuring that your request does not
include any confidential information that you or a third party may not
wish to be posted, such as medical information, your or anyone else's
Social Security number, or confidential business information, such as a
manufacturing process. The request for a hearing must include the
Docket No. FDA-2026-N-1321 for ``ChemoCentryx, Inc.; Proposal to
Withdraw Approval of New Drug Application for TAVNEOS (Avacopan)
Capsule, 10 Milligrams; Opportunity for a Hearing.'' The request for a
hearing will be placed in the docket and publicly viewable at <a href="https://www.regulations.gov">https://www.regulations.gov</a> or at the Dockets Management Staff between 9 a.m.
and 4 p.m., Monday through Friday. The applicant may submit all data
and analyses upon which the request for a hearing relies in the same
manner as the request for a hearing except as follows:
<bullet> Confidential Submissions--To submit any data analyses with
confidential information that you do not wish to be made publicly
available, submit your data and analyses only as a written/paper
submission. You should submit two copies total of all data and
analyses. One copy will include the information you claim to be
confidential with a heading or cover note that states ``THIS DOCUMENT
CONTAINS CONFIDENTIAL INFORMATION.'' The Agency will review this copy,
including the claimed confidential information, in its consideration of
any decisions on this matter. The second copy, which will have the
claimed confidential information redacted/blacked out, will be
available for public viewing and posted on <a href="https://www.regulations.gov">https://www.regulations.gov</a>
or available at the Dockets Management Staff between 9 a.m. and 4 p.m.,
Monday through Friday. Submit both copies to the Dockets Management
Staff. Any information marked as ``confidential'' will not be disclosed
except in accordance with 21 CFR 10.20 and other applicable disclosure
law.
Comments Submitted by Other Interested Parties: For all comments
submitted by other interested parties, submit comments as follows:

Electronic Submissions

Submit electronic comments in the following way:
<bullet> Federal eRulemaking Portal: <a href="https://www.regulations.gov">https://www.regulations.gov</a>.
Follow the instructions for submitting comments. Comments submitted
electronically, including attachments, to <a href="https://www.regulations.gov">https://www.regulations.gov</a>
will be posted to

[[Page 23279]]

the docket unchanged. Because your comment will be made public, you are
solely responsible for ensuring that your comment does not include any
confidential information that you or a third party may not wish to be
posted, such as medical information, your or anyone else's Social
Security number, or confidential business information, such as a
manufacturing process. Please note that if you include your name,
contact information, or other information that identifies you in the
body of your comments, that information will be posted on <a href="https://www.regulations.gov">https://www.regulations.gov</a>.
<bullet> If you want to submit a comment with confidential
information that you do not wish to be made available to the public,
submit the comment as a written/paper submission and in the manner
detailed (see ``Written/Paper Submissions'' and ``Instructions'').

Written/Paper Submissions

Submit written/paper submissions as follows:
<bullet> Mail/Hand Delivery/Courier (for written/paper
submissions): Dockets Management Staff (HFA-305), Food and Drug
Administration, 5630 Fishers Lane, Rm. 1061, Rockville, MD 20852.
<bullet> For written/paper comments submitted to the Dockets
Management Staff, FDA will post your comment, as well as any
attachments, except for information submitted, marked and identified,
as confidential, if submitted as detailed in ``Instructions.''
Instructions: All submissions received must include the Docket No.
FDA-2026-N-1321 for ``ChemoCentryx, Inc.; Proposal to Withdraw Approval
of New Drug Application for TAVNEOS (Avacopan) Capsule, 10 Milligrams;
Opportunity for a Hearing.'' Received comments, those filed in a timely
manner (see ADDRESSES), will be placed in the docket and, except for
those submitted as ``Confidential Submissions,'' publicly viewable at
<a href="https://www.regulations.gov">https://www.regulations.gov</a> or at the Dockets Management Staff between
9 a.m. and 4 p.m., Monday through Friday, 240-402-7500.
<bullet> Confidential Submissions--To submit a comment with
confidential information that you do not wish to be made publicly
available, submit your comments only as a written/paper submission. You
should submit two copies total. One copy will include the information
you claim to be confidential with a heading or cover note that states
``THIS DOCUMENT CONTAINS CONFIDENTIAL INFORMATION.'' The Agency will
review this copy, including the claimed confidential information, in
its consideration of comments. The second copy, which will have the
claimed confidential information redacted/blacked out, will be
available for public viewing and posted on <a href="https://www.regulations.gov">https://www.regulations.gov</a>.
Submit both copies to the Dockets Management Staff. If you do not wish
your name and contact information to be made publicly available, you
can provide this information on the cover sheet and not in the body of
your comments and you must identify this information as
``confidential.'' Any information marked as ``confidential'' will not
be disclosed except in accordance with 21 CFR 10.20 and other
applicable disclosure law. For more information about FDA's posting of
comments to public dockets, see 80 FR 56469, September 18, 2015, or
access the information at: <a href="https://www.govinfo.gov/content/pkg/FR-2015-09-18/pdf/2015-23389.pdf">https://www.govinfo.gov/content/pkg/FR-2015-09-18/pdf/2015-23389.pdf</a>.
Docket: For access to the docket to read background documents or
the electronic and written/paper comments received, go to <a href="https://www.regulations.gov">https://www.regulations.gov</a> and insert the docket number, found in brackets in
the heading of this document, into the ``Search'' box and follow the
prompts and/or go to the Dockets Management Staff, 5630 Fishers Lane,
Rm. 1061, Rockville, MD 20852, 240-402-7500.

FOR FURTHER INFORMATION CONTACT: Joan Dailey, Center for Drug
Evaluation and Research, Food and Drug Administration, 10903 New
Hampshire Ave., Bldg. 51, Rm. 6248, Silver Spring, MD 20993, 301-796-
6357, <a href="/cdn-cgi/l/email-protection#b4dedbd5da9ad0d5ddd8d1cdf4d2d0d59adcdcc79ad3dbc2"><span class="__cf_email__" data-cfemail="cca6a3ada2e2a8ada5a0a9b58caaa8ade2a4a4bfe2aba3ba">[email&#160;protected]</span></a>.

SUPPLEMENTARY INFORMATION:

I. Proposal to Withdraw Approval of NDA 214487

A. Summary

CDER proposes to withdraw approval of TAVNEOS (avacopan), NDA
214487, because of new information, which was withheld from FDA and did
not become known to FDA until more than three years after approval,
indicating that there is a lack of substantial evidence of
effectiveness for the drug, and the application contains untrue
statements of material facts.
Section 505(e) of the Federal Food, Drug, and Cosmetic Act (FD&C
Act), 21 U.S.C. 355(e), states that the Secretary of Health and Human
Services shall withdraw approval of a drug on certain grounds after due
notice and opportunity for hearing to the applicant. Under section
505(e)(3) of the FD&C Act and 21 CFR 314.150(a)(2)(iii), approval of a
drug shall be withdrawn if, on the basis of new information before FDA
with respect to the drug, evaluated together with the evidence
available when the application was approved, there is a lack of
substantial evidence from adequate and well-controlled investigations
that the drug will have the effect it purports or is represented to
have under the conditions of use prescribed, recommended, or suggested
in the labeling. In addition, under section 505(e)(5) of the FD&C Act
and 21 CFR 314.150(a)(2)(iv), approval of a drug shall be withdrawn if
FDA finds that the application contains any untrue statement of a
material fact.
New information shows that the applicant's unblinded study
personnel manipulated endpoint results for the phase 3 study,
CL010_168, referred to as the ADVOCATE study, which was the sole study
used to establish substantial evidence of effectiveness for approval of
TAVNEOS.\1\ That manipulation was designed to change results that were
not statistically significant and make the product look effective when
the original analysis did not support that conclusion. If the data as
originally analyzed according to the prespecified statistical analysis
plan had been submitted to the Agency, the study would not have been
viewed as establishing substantial evidence of effectiveness, which is
a legal requirement for approval of an NDA. However, in contravention
of fundamental principles of good clinical practice regarding the
reporting of clinical results as well as FDA's regulations regarding
what must be submitted in an NDA, the applicant did not submit this
data and original analysis to the Agency. Based on the new information
regarding data manipulation, along with the evidence available to CDER
when the application was approved, there is a lack of substantial
evidence that TAVNEOS will have the effect it purports or is
represented to have under the conditions of use prescribed,
recommended, or suggested in its labeling. Accordingly, CDER is
proposing to withdraw approval of TAVNEOS under section 505(e)(3) of
the FD&C Act and 21 CFR 314.150(a)(2)(iii).
---------------------------------------------------------------------------

\1\ See <a href="http://ClinicalTrials.gov">ClinicalTrials.gov</a> number NCT02994927, available at
<a href="https://clinicaltrials.gov/study/NCT02994927#more-information">https://clinicaltrials.gov/study/NCT02994927#more-information</a>
(accessed Apr. 24, 2026); see also Ref. 1.
---------------------------------------------------------------------------

We note that hepatotoxicity was identified as a key safety risk in
the original NDA review. In addition, serious and sometimes fatal cases
of hepatotoxicity have been reported postmarketing, and recent post-
market safety reports suggest that avacopan may also be associated with
vanishing bile duct syndrome (VBDS), a serious and, at the time of
approval, unexpected

[[Page 23280]]

adverse event. In light of our finding that there is a lack of
substantial evidence of efficacy, there is no demonstrated benefit to
balance the risks associated with TAVNEOS.
In addition, new information shows that the NDA for TAVNEOS
contains untrue statements of material fact, including statements
regarding when and how many times database lock and unblinding occurred
and inaccurate representations regarding the analyses that were used to
evaluate the ADVOCATE study. Because of these untrue statements, FDA
was unaware that the ADVOCATE study's primary endpoint data could not
be reliably interpreted as part of NDA approval and therefore the data
did not establish substantial evidence of effectiveness, which must be
demonstrated for approval under section 505(d) of the FD&C Act.
Instead, the approval of TAVNEOS was based on untrue statements
regarding the analyses of manipulated data, which is unreliable for
approval. Therefore, these untrue statements are material; as a result,
CDER is proposing to withdraw approval of TAVNEOS under section
505(e)(5) of the FD&C Act and 21 CFR 314.150(a)(2)(iv).

B. General Background

On October 7, 2021, FDA approved NDA 214487 for TAVNEOS (avacopan)
capsule, 10 mg, for the adjunctive treatment of adult patients with
severe active anti-neutrophil cytoplasmic autoantibody (ANCA)-
associated vasculitis (AAV) in combination with standard therapy
including glucocorticoids. The drug is included as a treatment option
in recent medical practice guidelines for AAV treatment. Standard
therapies for AAV include rituximab, cyclophosphamide, glucocorticoids,
and other immunosuppressants such as methotrexate and azathioprine.
The CDER review team identified a variety of issues that might
preclude approval of the product during the initial NDA review.
Following public discussion of these issues at a May 2021 advisory
committee meeting (Ref. 2), a shareholder lawsuit was filed that same
month against ChemoCentryx alleging securities fraud. In 2022, Amgen,
Inc. (Amgen) acquired TAVNEOS with its purchase of ChemoCentryx.
On May 29, 2025, plaintiffs in the securities fraud litigation
filed with the court the expert report from Marc Walton, M.D., Ph.D.
(Walton Report), (Ref. 3). The Walton Report stated that a statistical
analysis of the ADVOCATE study was conducted after database lock and
that the initial analysis had found that the avacopan treatment group
did not achieve a statistically significant superiority outcome on the
primary endpoint of sustained remission compared with the control
group. The Walton Report stated that, upon reviewing these results,
unblinded ChemoCentryx personnel had selected nine ADVOCATE study
subjects for readjudication following the initial blinded analysis,
including five avacopan patients whose clinical outcome would be
changed from ``not in sustained remission'' to ``sustained remission.''
The Walton Report further stated that, before sending the patient data
for readjudication, ChemoCentryx personnel confirmed that changing the
remission outcomes for the five avacopan patients would change the
study results to reflect that avacopan demonstrated statistically
significant superiority over therapy in the control arm. Finally, the
Walton Report stated that the five avacopan patients were readjudicated
to ``sustained remission,'' and the revised statistical analysis was
submitted to FDA in the NDA without disclosing this readjudication of
the pivotal study results. Specifically, the Walton Report noted that
the applicant did not disclose the original statistical analysis of the
study, the unblinded selection of subjects for readjudication after the
clinical database was locked, or that the post-database lock changes
changed the study result from not statistically significant to
statistically significant.
After learning about the Walton Report, CDER sent an information
request on July 23, 2025, seeking detailed information regarding the
handling of unblinded data in the ADVOCATE study. On August 22, 2025,
Amgen responded to the information request on behalf of the applicant
(Ref. 4). Amgen's response confirmed the key factual allegations
outlined above from the Walton Report. Amgen nevertheless claimed in
its response that the data in the NDA are accurate and that the
readjudications were appropriate. As explained below, FDA disagrees.
On August 15, 2025, the court dismissed the shareholders' lawsuit,
granting summary judgment in favor of ChemoCentryx. The court's opinion
did not acknowledge or address the data manipulation allegations in the
Walton Report and does not have any bearing on the status of the NDA or
enforcement of the FD&C Act.
This Notice of Opportunity for a Hearing omits certain information
that is not publicly available at this time.

C. ADVOCATE Study and Data Manipulation

1. Study Background
The ADVOCATE study randomized 331 subjects with AAV to avacopan 30
mg twice daily or a protocol-specified prednisone taper in a 1:1 ratio;
all subjects were treated with background immunosuppressive therapy of
rituximab or cyclophosphamide. Because glucocorticoids are used to
treat AAV, minimizing their use by study subjects would ensure that
efficacy outcomes were interpretable and not inflated. The study
protocol (Ref. 5), statistical analysis plan (Ref. 6), and clinical
study report (Ref. 7) state that glucocorticoids not supplied as a
study drug must be avoided ``as much as possible during the study.''
\2\ According to the protocol, study subjects were permitted to receive
``low doses'' of oral glucocorticoids (no more than 10mg/day) for
treatment of adrenal insufficiency or for other conditions (Ref. 5 at
124). Non-study supplied glucocorticoids were permitted only in low
doses for adrenal insufficiency, but this rule was liberalized over
time to also permit glucocorticoid use at any dose by study subjects
for conditions other than adrenal insufficiency (e.g., allergic
reaction) if they were not prescribed for treatment of AAV.\3\ The
protocol and statistical analysis plan are each internally inconsistent
on this point; on the one hand, they require non-study supplied
glucocorticoids to be avoided as much as possible, but on the other
hand permit such glucocorticoid use at any dose with no impact on
remission determinations.
---------------------------------------------------------------------------

\2\ Ref. 5 at 17, 50, and 123; Ref. 6 at 26; Ref. 7 at 31, 35.
\3\ See Ref. 8 at 8.
---------------------------------------------------------------------------

Patients were followed for 52 weeks for a disease remission
endpoint evaluated using the Birmingham Vasculitis Activity Score
(BVAS), a standardized, clinician reported outcome measure of disease
activity in patients with vasculitis. The primary endpoints for the
ADVOCATE study were remission at week 26 and sustained remission at
week 52. According to the study protocol, study subjects were
considered in remission at weeks 26 and 52 if they had a BVAS of zero
and had not received glucocorticoids for treatment of AAV within 4
weeks prior to the week 26 or week 52 assessment, respectively.
Sustained remission was defined as remission at weeks 26 and 52 with no
intervening relapse. The study protocol and statistical analysis plan
set forth a prespecified method for handling

[[Page 23281]]

missing data, under which subjects whose remission or sustained
remission status was not evaluated within a specified window of time
would be imputed as not in remission (week 26) or sustained remission
(week 52) (Ref. 4 at 6; Ref. 6 at Sec. Sec. 5.4, 5.5).
BVAS, remission at week 26, and sustained remission at week 52 were
determined by a blinded adjudication committee. The adjudication
committee's activities were governed by an adjudication committee
charter (AC Charter). The AC Charter prohibited readjudication of
remission status after database unblinding (Ref. 8, Addendum 2 Sec.
4.3.3).
ChemoCentryx originally proposed a study design in which the
primary endpoint of sustained remission at week 52 would be analyzed
for non-inferiority compared to glucocorticoids. FDA informed
ChemoCentryx in 2016 that TAVNEOS would not be considered for approval
unless the ADVOCATE study demonstrated superiority on week 52 sustained
remission results in the avacopan arm of the study compared to the
control arm (Ref. 9 at 2, 4).
2. Data Manipulation
FDA takes seriously allegations of scientific misconduct that
threaten the integrity of the drug approval process and patient safety.
Upon reviewing the Walton Report, FDA became concerned that the data
submitted to the NDA were biased and that the ADVOCATE study did not
provide reliable evidence of the effectiveness of TAVNEOS. Accordingly,
FDA requested that the applicant provide a comprehensive account of the
events described in the Walton Report and the handling of unblinded
adjudication data in the ADVOCATE study. FDA also requested emails, raw
data, and other documents relevant to the data manipulation described
in the Walton Report.
Amgen's response to CDER's July 2025 information request (Amgen
Response) chronicled the facts regarding the handling of data in the
ADVOCATE study. The Amgen Response states that the contract research
organization, Medpace, Inc. (Medpace), conducted prespecified
procedures to ``clean'' the ADVOCATE study database to ensure the data,
including the primary endpoint data, were accurate and complete before
the database was locked for analysis.\4\ Specifically, on November 4,
2019, the Medpace lead biostatistician completed a biostatistics pre-
database lock checklist verifying, among other things, that the
database and datasets were ready for analysis, the statistical analysis
plan had been approved, and the statistical programs had been validated
(Ref. 4, Exh. 3). On November 5, 2019, the Medpace data coordinator and
data manager signed a database lock checklist, confirming among other
things that all clinical data had been entered, all edit checks
applied, all queries resolved, source data verified, all records locked
from editing, and all other necessary steps completed as a prerequisite
for the database to be declared clean and ready to be locked (Ref. 4,
Exh. 4). The study database was then locked and unblinded on November
5, 2019.
---------------------------------------------------------------------------

\4\ See Ref. 4 at 7-8. Database cleaning addresses problems with
incomplete, invalid, or inconsistent data caused by improper data
recording or data entry (Ref. 10). Database cleaning activities may
include reconciliation of entered data and datasets, rectification
of data errors, and addressing the impact of noncompliance issues,
including protocol deviations. Ref. 11 at 47. The sponsor should
ensure that blinding is maintained during the process of cleaning
the database prior to its release for analysis (Ref. 12 at 11).
---------------------------------------------------------------------------

Between November 5, 2019, and November 7, 2019, Medpace took a
database snapshot and generated datasets and tables, listings, and
figures (TLFs) for unblinded review and validation by select
ChemoCentryx personnel (Ref. 4 at 10). With respect to efficacy data,
the Study Results Analysis Plan (SRAP), which was signed on November 4,
2019, permitted two ChemoCentryx employees to receive unblinded data
after database lock for the purposes of conducting a quality control
(QC) review ``for completeness and internal consistency'': Dr. Huibin
Yue, Director of Biostatistics, and Dr. Pirow Bekker, Chief Medical
Officer (Ref. 4 at 9-10 and Exh. 2). In addition, Dr. Chao Wang, an
independent statistical consultant retained by ChemoCentryx, was
responsible for assisting in the validation of topline TLFs (Ref. 4 at
10).
On November 8, 2019, in accordance with the SRAP, the unblinded
datasets and top-line efficacy TLFs were transferred to Dr. Bekker.
Those results showed that the primary endpoint of sustained remission
at week 52 had failed to achieve statistical significance, returning a
two-sided p-value of 0.1025, which is above the significance threshold
specified in the statistical analysis plan.\5\ The applicant did not
include these results in the NDA or otherwise disclose them to FDA
during the Agency's review of the application.
---------------------------------------------------------------------------

\5\ Ref. 13. The statistical analysis plan specified a procedure
to preserve the overall Type I error rate at a 5% level with a one-
sided p-value threshold of 0.025 (Ref. 6 at 30-31). For equivalent
comparison to a two-sided 5% significance level (i.e., two-sided p-
value threshold of 0.05) and alignment with current labeling, two-
sided p-values are presented in this document.
---------------------------------------------------------------------------

On November 9, 2019, Dr. Bekker emailed Dr. Yue requesting that Dr.
Yue and Dr. Wang verify the week 52 sustained remission results:
``Regarding sustained remission at Week 52 it is, of course, of
paramount importance that the data are correct (especially the p-value
for superiority). Chao and you must verify these results ASAP. We
cannot afford to miss a superiority outcome here'' (Ref. 4, Exh. 11
(emphasis in original)). On November 11, 2019, Dr. Yue emailed the
Medpace lead biostatistician to request a blinded data quality check to
confirm whether all adjudicated remission determinations in the
electronic data capture (EDC) system match those on paper case report
forms (CRFs). His email noted that ``[w]e cannot afford to miss a
superiority outcome,'' that ``[t]he p-value is close to be[ing]
significant,'' and that ``[w]e need to make sure that both the data and
analysis are correct'' (Ref. 4, Exh. 12). The email further noted that
Dr. Yue was awaiting Dr. Wang's validation of the statistical analysis.
The lead biostatistician replied by email, dated November 12, 2019,
that the quality check had been completed ``with no findings.'' Dr.
Wang validated the TLF analysis on November 12, 2019.
After being informed that there was nothing wrong with either the
data or the analysis, Drs. Bekker and Yue looked for cases that could
alter the study results to indicate TAVNEOS achieved a statistically
superior outcome on the primary endpoint of sustained remission. On
November 12, 2019, Dr. Yue ran a targeted, unblinded search of the
clinical database to identify all subjects with a BVAS of zero who used
glucocorticoids in the four weeks prior to the week 26 or week 52
remission assessments and who were originally adjudicated as not in
remission (Ref. 4, Exh. 14). He forwarded the resulting unblinded data
spreadsheet to Dr. Bekker. In addition, Drs. Yue and Bekker reviewed
unblinded data to determine whether data from early termination or
unscheduled visits was included in the primary endpoint analysis
consistent with the statistical analysis plan (Ref. 4 at 12; Ref. 4,
Exh. 13). Together, they identified nine subjects--three from the
prednisone arm and six from the avacopan arm--for readjudication. Most
were identified for readjudication because Drs. Bekker and Yue believed
their primary endpoints were inconsistent with the prespecified
protocol provisions related to glucocorticoid use or missing data, or
both. One subject was identified for

[[Page 23282]]

readjudication due to a manual data entry error in which the subject
was categorized as not in remission or sustained remission despite
satisfying the relevant criteria.
On November 13, 2019, Dr. Bekker forwarded to Dr. Yue a table
identifying the nine subjects, their previously adjudicated remission
determinations, and the rationale for a potential revised adjudication
decision (Ref. 4, Exh. 15). That same day, Dr. Yue asked Dr. Wang to
rerun the primary endpoint analyses and calculate new p-values based on
a hypothetical dataset accounting for the anticipated results of the
readjudication of the nine subjects. Specifically, the dataset
anticipated that five of the six subjects in the avacopan arm would be
readjudicated to sustained remission, and none of the three subjects in
the control arm would be readjudicated to sustained remission. On
November 13, 2019, Dr. Wang replied that the new p-value showed
statistical significance.
On November 14, 2019, Dr. Yue forwarded to Medpace personnel the
table identifying the subjects for potential readjudication. Later the
same day, Dr. Bekker spoke with the ADVOCATE study's Adjudication
Committee (AC) Chair, Dr. David Jayne, regarding the readjudications
(Ref. 4, Exh. 18). On November 14, 2019, after receiving confirmation
from Dr. Bekker that Dr. Jayne was expecting cases for readjudication,
Medpace sent Dr. Jayne a data package (``patient profile'') for each of
the nine subjects to be readjudicated. Each patient profile included
the subject's original adjudication form, BVAS and other remission-
related data, and the same table Dr. Bekker forwarded to Dr. Yue the
previous day, minus the rationale for a potential revised adjudication
decision for each subject (Ref. 4, Exh. 18). The table directed Dr.
Jayne's attention to the specific facts for each subject that were
relevant to review (i.e., unscheduled visit within assessment window
and glucocorticoid use). In preparation for potential database changes
to record readjudicated remission determinations, Dr. Yue and Medpace
personnel signed a post-datalock change approval form on November 15,
2019 (Ref. 4, Exh. 69).
Dr. Jayne conducted a blinded readjudication of all nine subjects.
Among other reclassifications, Dr. Jayne reclassified five of the six
avacopan subjects as in sustained remission. Table 1 summarizes the
adjudication and readjudication decisions: \6\
---------------------------------------------------------------------------

\6\ The Amgen Response erroneously identified subject 852-001 as
in the treatment arm (Ref. 4 at 14). According to the ChemoCentryx
response to an information request submitted during NDA review, this
subject was in the control arm (Ref. 14 at 13). Table 1 reflects
readjudication decisions by Dr. Jayne and should not be interpreted
as FDA's agreement that these readjudication decisions were properly
determined.

Table 1--Effect of Readjudication
----------------------------------------------------------------------------------------------------------------
Before readjudication After readjudication*
----------------------------------------------------------------------------------------------------------------
Week 52 sustained Week 52 sustained
Treatment arm and subject ID Week 26 remission remission Week 26 remission remission
----------------------------------------------------------------------------------------------------------------
Control arm (prednisone):
428-001..................... No................ No (ET)........... Yes............... No (ET)
852-001..................... No................ No................ Yes............... No
957-002..................... No................ No (ET)........... Yes............... No (ET)
Treatment arm (avacopan):
440-002..................... No................ No................ Yes............... Yes
466-003..................... No................ No................ Yes............... Yes
534-001..................... No................ No................ Yes............... Yes
702-001..................... No................ No................ Yes............... Yes
751-001..................... Yes............... No................ Yes............... Yes
854-001..................... No................ No................ Yes............... No
----------------------------------------------------------------------------------------------------------------
ET: early termination; remission at week 52 imputed as nonresponse.
*Bold text in the two columns under the ``After Readjudication'' heading indicates readjudicated determinations;
unbolded text in the same two columns indicates no readjudication was requested (e.g., because of early
termination or excessive glucocorticoid use).

The endpoint determinations were changed in the database
accordingly, and the database was locked for a second time on November
20, 2019. At that point, the primary efficacy analysis was rerun, and
this time generated statistically significant results. These were the
only primary endpoint analysis results submitted in the NDA to support
the demonstration of substantial evidence of effectiveness for
approval.
3. Study Documents Prohibited Readjudications After Database Lock
None of the nine readjudications was justified or permitted under
the terms of the AC Charter. According to the AC Charter,
readjudications should be rare. Specifically, the AC Charter states
that if an adjudication form is complete, the remission time point will
be considered adjudicated and ``[t]he form will not be presented to the
AC again at a later time'' (Ref. 8, Sec. 4.3.2). This makes sense
because all data packages presented to the AC must be ``100% cleaned,''
having relevant data sources verified, manual/auto edits run, and
queries resolved (Ref. 8, Sec. 4.1, Addendum 2, Sec. 4.3.3). Under
the terms of the AC Charter, readjudication should occur only when the
original adjudication decision could be affected by data changes made
by investigative sites after adjudication, such as changes to BVAS or
other data relevant to remission status (Ref. 8, Addendum 2, Sec.
4.3.3). Medpace was responsible for tracking post-adjudication data
changes and providing the AC Chair or designee with periodic listings
of data changes (Ref. 8, Addendum 2, Sec. 4.3.3). In this case, there
were no underlying ``data changes'' that precipitated this request for
readjudication--only a desire for changes that would alter the study
results to indicate avacopan was effective when the original analysis
of the study failed to find such an effect.
Moreover, the AC Charter indicates that no readjudications can
occur after unblinding--a key provision to prevent bias in the study
conduct, analysis, and results. Specifically, the AC Charter states
that ``no changes may be entered into EDC after data base lock'' (Ref.
8, Addendum 2, Sec. 4.3.3). EDC was the electronic data capture system
used to enter adjudication decisions (Ref. 8, Sec. 4.3). The AC
Charter further states that ``[t]here will be no blinded BVAS listings
for review after unblinding''

[[Page 23283]]

(Ref. 8, Addendum 2, Sec. 4.3.3).\7\ Although this provision appears
limited to BVAS listings, it essentially prohibits any readjudication
after database unblinding because every determination of remission
status involves an assessment of BVAS data, among other factors.
Indeed, every patient profile sent for adjudication must include not
only BVAS listings, but also a variety of other information relevant to
remission status, including prior and concomitant medications
(including ``non-study-supplied glucocorticoid medications''), relapse
information, and previously adjudicated BVAS, VDI, remission, and
relapse data, if applicable (Ref. 8, Sec. 4.1.1). Thus, sending
blinded BVAS listings and other remission-related data to Dr. Jayne for
readjudication after the database had been unblinded violated the terms
of the AC Charter and therefore went beyond the ``QC review for
completeness and accuracy'' permitted under the SRAP after database
lock (Ref. 4, Exh. 2).
---------------------------------------------------------------------------

\7\ Similar language appears in the study protocol, which states
that ``BVAS data recorded by investigators will be adjudicated,
according to an adjudication charter, before data finalization and
unblinding. The adjudicated data will be used in the final
analysis'' (Ref. 5, Sec. 7.1.1).
---------------------------------------------------------------------------

The time to correct errors in study data is before unblinding,
which eliminates the potential to introduce bias in the data and
analysis as a result of knowing the assignment of subjects to the
control or treatment arms of a study. Before the ADVOCATE study
database was locked and unblinded on November 5, 2019, ChemoCentryx had
many opportunities to correct errors in study data, including those
involving the nine subjects identified for readjudication after
unblinding. For example, the purported incorrect application of the
prespecified provisions regarding missing data and glucocorticoid use
should have been identified and corrected as part of the patient
profile data cleaning conducted before the initial adjudication of any
study subject. In addition, any purported misapplication of the study
provisions regarding glucocorticoid use or missing data should have
been caught over the course of the trial and certainly during database
cleanup prior to unblinding.
Another opportunity for correcting errors, including those
involving remission status for subjects with missing data, was the
blinded data review meeting held on October 24, 2019, with Medpace and
ChemoCentryx staff, including Drs. Bekker and Yue.\8\ Among other
things, the purpose of the meeting was to identify significant protocol
deviations that would require excluding subjects from the per protocol
study analysis. At this meeting, Dr. Bekker identified four subjects
for exclusion from the per protocol population ``because their response
could not be assessed at week 26 and are therefore non-responders.''
(Ref. 4, Exh. 56). After unblinding, Drs. Bekker and Yue identified one
of the four subjects--avacopan subject 534-001--for readjudication on
the grounds that the prespecified method for handling missing data had
not been applied accurately. Prior to unblinding Dr. Bekker was aware
of the rules for handling missing data and concluded that this subject
was a non-responder. Only after unblinding did Dr. Bekker take a
different position so that the subject could be readjudicated. That
subject was readjudicated as being in remission, which contributed to
changing the study outcome. This purported error--as well as any other
errors involving misapplication of the prespecified method for handling
missing data--could have been identified before or at the blinded data
review meeting, or during the QC reviews performed prior to database
lock and unblinding.
---------------------------------------------------------------------------

\8\ Ref. 4, Exh. 56. The SAP required this meeting to be held
prior to database lock and unblinding (see Ref. 6 at 24).
---------------------------------------------------------------------------

Finally, as required by the AC Charter, BVAS and other data related
to remission, as well as ``[t]he 52 week adjudications,'' should have
been cleaned before database lock, which occurred on November 5, 2019.
That was the applicant's last chance to correct adjudication decisions
because the AC Charter prohibited changes to adjudication decisions
after database lock (Ref. 8, Addendum 2, Sec. 4.3.3). Other data
errors or inconsistencies, such as the manual data entry error
involving one of the readjudicated subjects (subject 466-003), should
have been identified during the various QC checks conducted immediately
before database lock and unblinding, if not earlier.
The applicant's conduct cannot be considered ``quality control'';
rather, sponsor personnel engaged in data manipulation that distorted
the study results as reported in the NDA, which were material to the
review of the NDA. As more fully explained in section I.E.1 below, the
conduct biased the study results reported in the NDA to the extent that
the pivotal study cannot be relied upon to support a demonstration of
substantial evidence of effectiveness.

D. NDA Review Based on Manipulated Primary Endpoint Results

In the NDA submission, the applicant represented that the ADVOCATE
study demonstrated (1) superiority for sustained remission at week 52
(avacopan versus prednisone treatment difference 12.5 percent with 95
percent confidence interval (CI) [2.6 percent, 22.3 percent] and two-
sided p-value = 0.0132 for superiority at week 52); and (2)
noninferiority, but not superiority, for remission at week 26 (avacopan
versus prednisone treatment difference 3.4 percent with 95 percent CI
[-6.0 percent, 12.8 percent], two-sided p-value < 0.0001 for
noninferiority at week 26).
The CDER review team sent several information requests to the
applicant during the NDA review to better understand the adjudication
process and results. However, the applicant never disclosed the
readjudications that occurred after the initial database lock and that
they were prompted by both an unblinded review of the study database
and knowledge that the readjudications would change the study result
from not statistically significant to statistically significant. In a
December 9, 2020, information request, CDER inquired about ``document
finalization and unblinding dates'' in an effort to establish whether
the study's statistical analysis plan had been finalized before data
unblinding (Ref. 15). Noting that the specified data analyses in the
statistical analysis plan differed from those in the original study
protocol, CDER expressed concern that perhaps the choice of analyses
specified in the statistical analysis plan was influenced by viewing
trial data after unblinding, which would have introduced significant
bias and inflated Type 1 error (i.e., the chance of a false positive,
which is a finding of a significant difference between drug and control
when there really is none). The applicant's response indicated that the
statistical analysis plan was finalized before the study database was
frozen (i.e., locked) on November 20, 2019 (Ref. 16). The response
falsely characterized this as the only date on which the database was
locked, because it refers to ``the freeze of the study database, which
occurred on 20 November 2019.'' ChemoCentryx omitted any information
regarding the database lock and unblinding that occurred on November 5,
2019. Moreover, the applicant's response omitted the crucial point that
its own personnel reviewed unblinded primary efficacy endpoint data
before the November 20, 2019, database lock. Nor did the applicant
disclose that its unblinded personnel caused the readjudication of
primary endpoint data for nine study subjects, which ultimately changed
the result from not

[[Page 23284]]

statistically significant to statistically significant. At no point did
the applicant disclose any information about the first database lock,
including its failed topline efficacy results, and that nine patients'
endpoint determinations were readjudicated in response to a request
from unblinded study personnel.
The applicant's failure to disclose these facts is contrary to
well-established internationally recognized good clinical practice
(GCP) guidelines established for the conduct of clinical trials,
including principles for appropriate statistical analysis. These
principles serve as a foundation for helping to ensure that data and
statistical analysis are free from biases that make the study results
uninterpretable (Ref. 11 at 42, Ref. 12 at 10-12). FDA has reiterated
these principles in several FDA guidance documents. For example, FDA
guidance states that changes made to data after unblinding should occur
only in exceptional circumstances, should be clearly documented and
justified, and reported in the clinical study report (CSR).\9\ The CSR
submitted to the NDA for the ADVOCATE trial made it seem like the
applicant followed accepted statistical analysis practices when they
did not. Specifically, the CSR did not disclose that adjudication data
were changed after the database was initially locked and after
unblinding, nor did it disclose that the post-unblinding readjudication
of primary endpoint data reversed the failed study results in the
original analysis (Ref. 7).
---------------------------------------------------------------------------

\9\ Ref. 11 at 42. ``Exceptional circumstances'' may include
``discrepancies that must be resolved for the reliability of the
trial results.'' Id. Here, other than the correction of a manual
error, the errors to be corrected were not mere data
``discrepancies,'' such as transcription errors or discrepancies
between source documents and case report forms recorded in the study
database. In addition, the manner in which the errors were resolved
here--which was contrary to the express terms of the AC Charter and
not disclosed to FDA--only reduced the reliability of the study
results.
---------------------------------------------------------------------------

Documents submitted to the NDA describe the November 20, 2019,
database lock and resulting unblinding and statistical analysis as if
it were the only one that occurred. For example, in the previously
described response to CDER's December 9, 2020 information request, the
applicant characterized the study database lock as a single event:
``the freeze of the study database, which occurred on 20 November
2019'' (Ref. 16 at 1). The CSR states that ``[t]he database was frozen
on November 20, 2019'' to conduct the primary efficacy endpoint
analysis (Ref. 7 at 53), omitting the critical fact that the database
had been frozen earlier to conduct the original analysis. In contrast,
the CSR discloses elsewhere that ``minor'' errors regarding vasculitis
damage index (VDI) scores were discovered after database lock, and the
report includes original and revised listings and tables for VDI scores
(Ref. 7 at 64-65). The CSR further states that ``[t]reatment
assignments for individual subjects remained blinded to the study team,
Investigators, and subjects until after the study database had been
cleaned and locked'' (Ref. 7 at 34). That is untrue with respect to the
November 20, 2019, database lock date, which was represented to FDA to
be the sole database lock date.
Protecting the study blinding is an expected element of proper
statistical planning and analysis. Indeed, the study documents
recognize the importance of maintaining the study blinding before the
database was locked for analysis. For example, the statistical analysis
plan set forth safeguards against unblinding of study data ``prior to
database freeze.'' The applicant appears to have followed these
provisions of the statistical analysis plan and study protocol when it
locked the study database on November 5th and then shared unblinded
data with applicant staff. However, the applicant violated the
statistical analysis plan and acceptable statistical practices once it
decided to unlock the database, use unblinded data to request that nine
patients be readjudicated in violation of the adjudication committee
charter, reanalyze the data post readjudication, and submit those
results to FDA as though the first database lock and analysis had never
occurred.
Before approval, whether the applicant submitted sufficient
evidence in its NDA to demonstrate TAVNEOS had its intended effect was
debatable across the FDA reviewers and independent advisory committee
members. The 18-member advisory committee FDA convened to provide
expert advice regarding the evidence submitted in the TAVNEOS NDA were
split about whether the data were sufficient to demonstrate that the
product was effective, safe, or that the benefits outweighed the risks
(Ref. 17). Moreover, after considering the application, including the
applicant's responses to information requests, the study protocol,
statistical analysis plan, and clinical study report, the primary
review team did not recommend approval of TAVNEOS. In other words, even
without knowledge of the data manipulation by unblinded personnel, the
primary review team had significant concerns regarding whether there
was a demonstration of substantial evidence of effectiveness (Ref. 18
at 33).
Importantly, the purported statistically significant superiority
for sustained remission at week 52 was a key factor in the
recommendation to approve the NDA made by the Director of the Division
of Rheumatology & Transplant Medicine and the decision to approve the
NDA made by the Director of the Office of Immunology & Inflammation
(Ref. 18 at 33-35). In addition, the Division Director relied on what
he was led to believe were study data and results that reflected
appropriate adherence to the study plan documents. Specifically, the
Division Director believed that as a large, global, multicenter trial,
with procedures in place to ensure trial quality, the ADVOCATE study
would be less vulnerable to certain biases, such as selection or
measurement bias (Ref. 18 at 34). In recommending approval, the
Division Director also considered the clinical context, noting that AAV
is a rare disease with an unmet need for additional therapeutic options
and a need for treatment alternatives with fewer toxicities (Ref. 18 at
35). Under the circumstances, the Division Director believed additional
flexibility was warranted with respect to the acceptability of the
uncertainties highlighted by the primary review team, and the Office
Director concurred (Ref. 18 at 35). The Division Director did not know
that the facts on which he was relying were untrue--that absent
manipulation, the study results were not statistically significant on
the superiority analysis at week 52, and the procedures in place to
ensure trial quality were violated. Had the Division Director known
these facts, he would not have recommended approval. These facts and
the Division Director's recommendation would have been material to the
review of the application by the Office Director and her decision to
approve it.
Because of the limited safety data in the original NDA, the Agency
required a postmarketing study (PMR 4155-1) to evaluate safety
outcomes, including hepatotoxicity, DILI, and serious hypersensitivity
reactions (Ref. 19). The same study is also assessing efficacy outcomes
to fulfill a postmarketing commitment agreed to during NDA review (Ref.
18 at 231). According to the applicant's most recent annual report for
this ongoing study, only 21 of the planned 300 patients have been
enrolled.

E. Impact of Data Manipulation and Untrue Statements

Immediately upon learning that the results of the sole study
intended to support a demonstration of effectiveness

[[Page 23285]]

were not positive and knowing this would almost certainly result in a
failure to obtain approval, key personnel took multiple steps to
manipulate the results to change the study result from not
statistically significant to statistically significant and hide that
manipulation from the Agency by making untrue statements in the NDA.
Although Amgen states the readjudications corrected errors in the
study data, the applicant's actions were improper and rendered the
results biased and unreliable. No further analysis of the data can
rectify this problem. Adequate and well-controlled clinical
investigations are designed, among other things, to minimize the risk
of a false positive conclusion (Type 1 error).\10\ Statistical analyses
are prespecified and controlled to minimize this risk. When a study
entails multiple analyses of a primary endpoint, statistical procedures
are used to control the increased risk of false positives. Without such
procedures, the risk of false positives will increase. The ADVOCATE
study was not designed to control for the multiple unplanned hypothesis
tests for the primary endpoints that the applicant conducted.
Therefore, the Type 1 error risk was not properly controlled, and the
primary endpoint of sustained remission at 52 weeks presented in the
NDA cannot be reliably interpreted.
---------------------------------------------------------------------------

\10\ See 21 CFR 314.126.
---------------------------------------------------------------------------

The ADVOCATE study cannot be considered adequate and well-
controlled. For example, FDA regulations at 21 CFR 314.126(b)(5)
require ``an analysis of the results of the study adequate to assess
the effects of the drug.'' The applicant conducted a reanalysis of the
data that did not control for multiplicity, and therefore the analysis
could not adequately assess the effects of the drug. Moreover, under
314.126(b)(5), a study must ensure that ``adequate measures are taken
to minimize bias on the part of the subjects, observers, and analysts
of the data.'' After the applicant's analysts knew that the study
failed to achieve statistical significance, it used unblinded data to
select certain subjects for readjudication with the goal of changing
the statistical significance of the study outcome. Therefore, the study
conduct did not adequately minimize bias on the part of the analysts of
the data. Accordingly, the ADVOCATE study cannot support a
demonstration of substantial evidence of effectiveness. The only
interpretable analysis of the primary endpoint of sustained remission
at 52 weeks was the original analysis based on the November 5, 2019,
database lock, which was conducted in accordance with the prespecified
analysis plan after the data were checked and confirmed to be accurate
before unblinding. That analysis was not statistically significant and
would not have supported a demonstration of substantial evidence of
effectiveness during the NDA review.
1. Impact of the Data Manipulation
The applicant's manipulation of data in the ADVOCATE study
irreconcilably biased the study result on sustained remission and
renders that result uninterpretable for regulatory decision making.
Minimizing bias in clinical research is essential to ensure that study
results are reliable and reflective of a medical product's safety and
effectiveness. Bias can arise in many ways, including through clinical
trial design and conduct, as well as the analysis and evaluation of
study results.
Blinding is one of the most important design techniques for
avoiding bias in clinical trials (Ref. 12 at 10). Blinding is intended
to limit the occurrence of conscious and unconscious bias in the
conduct and interpretation of a clinical trial arising from the
influence that the knowledge of treatment may have on a number of
factors, including the assessment of endpoints, the exclusion of data
from analysis, and more (Ref. 12 at 10). Maintaining the blinding until
all opportunities for bias have passed is critical. Regardless of how
bias is introduced--whether consciously or unconsciously, from lack of
blinding, improper unblinding, selective outcome reporting, etc.--it
distorts findings and reduces the reliability of study results.
a. Biased Study Results
In its August 22, 2025 response to FDA, Amgen claimed that the
readjudications were statistically appropriate to preserve the accuracy
of the underlying trial data (Ref. 4 at 28). Amgen's claim is
incorrect. While it may be generally acceptable to conduct blinded data
quality assurance checks, the unblinded review of study data to inform
patient selection for readjudication and reanalyses--and the failure to
disclose it--was a significant deviation from core scientific and
statistical principles and good clinical practices for clinical
trials.\11\ In any double-blind study, such as the ADVOCATE study,
neither the subject nor any of the investigators or sponsor staff--
including those evaluating endpoints--should be aware of the treatment
received (Ref. 12 at 11). Moreover, this level of blinding must be
maintained throughout the conduct of the trial; ``only when the data
are cleaned to an acceptable level of quality will appropriate
personnel be unblinded'' (Ref. 12 at 11). FDA guidance further states
the fundamental principle that any intentional or unintentional
breaking of the blinding should be reported and explained to the
Agency, regardless of the reason for its occurrence (Ref. 12 at 12). In
addition, post-unblinding data changes and deviations from the planned
statistical analyses should be reported in the clinical study report
(Ref. 11 at 42), which is included in the NDA submission.\12\
---------------------------------------------------------------------------

\11\ See, for example, discussion of bias and the importance of
blinding in section II.C of the guidance for industry entitled ``E9
Statistical Principles for Clinical Trials'' (Ref. 12). See also the
guidance for industry entitled ``E6(R3) Good Clinical Practice''
(Ref. 11).
\12\ See 21 CFR 314.50(a) (requiring an NDA to ``contain reports
of all investigations of the drug product sponsored by the
applicant''); see also Ref. 20.
---------------------------------------------------------------------------

As discussed in section I.C.3 above, the ADVOCATE study had several
procedures designed to correct study data before database unblinding.
However, the correction of study data after unblinding and analysis can
introduce substantial conscious and unconscious bias. Here, the
evidence shows that the applicant and its service providers conducted a
quality control review of the data before the database was locked on
November 5, 2019 and identified no concerns.
What the applicant did is exactly what the AC Charter was designed
to prevent: the introduction of bias in the study results due to
selective readjudication of study subjects after unblinding. In the
absence of any evidence of erroneous data, sponsor personnel used
unblinded data specifically to find subjects whose readjudication would
change the results of the trial. The evidence shows that the goal of
this behavior was to alter the results of the original analysis. If the
original analysis had been positive, Drs. Bekker and Yue would likely
not have searched for subjects to readjudicate.
Knowledge of the study results permitted the applicant to conduct a
biased search for subjects to readjudicate where the readjudication
would change the study results from negative to positive. Although the
Amgen Response indicates that Drs. Bekker and Yue analyzed remission
data for all patients in the study, they appeared to focus primarily on
a subpopulation of patients who had a BVAS of zero and were adjudicated
as not in remission due to missing data or recent glucocorticoid use
for a reason

[[Page 23286]]

other than treatment of AAV (Ref. 4 at 11-14; Ref. 4, Exh. 14). This
targeted review would not detect misapplication of other prespecified
protocol provisions affecting remission determinations, such as the
definition of disease relapse or the accuracy of BVAS assessments, in
subjects that had been adjudicated as in remission or sustained
remission. Their targeted data review was biased in favor of finding
more patients--specifically more avacopan patients--whose remission
status could be changed to support the desired trial outcome.
In addition, knowledge of treatment assignments appears to have
biased which study subjects were referred for readjudication. We note
that the remission determinations selected for readjudication were
heavily weighted in favor of generating a statistically significant
result on sustained remission for subjects in the avacopan arm of the
ADVOCATE study. For example:
<bullet> All subjects identified for readjudication were believed
to have been incorrectly adjudicated as not in remission at week 26 or
week 52 or both;
<bullet> Twice as many subjects in the avacopan arm were selected
for readjudication compared to the control arm (six versus three); and
<bullet> The three control arm patients selected for readjudication
could not have made a difference in the analysis of the superiority
claim at week 52 because two of the three terminated too early to be
evaluated for sustained remission, and glucocorticoid use by the third
control subject precluded a sustained remission determination.\13\ In
fact, readjudication of the week 52 remission status for the three
control arm subjects was not even requested.\14\
---------------------------------------------------------------------------

\13\ See Ref. 4 at 12-14 regarding subjects 428-001, 957-002,
and 852-001. We note that subject 852-001 was erroneously
characterized in that document as being in the avacopan treatment
arm. This subject was in the control arm of the ADVOCATE study (Ref.
14 at 13).
\14\ See Ref. 4, Exh. 18.
---------------------------------------------------------------------------

Bias in the selection of subjects for readjudication could not be
mitigated by the fact that Dr. Jayne conducted the readjudications in a
blinded fashion. The AC charter rightly prohibits readjudication after
study unblinding because of the potential for bias. In addition, bias
is not overcome because the applicant also included three prednisone
control subjects for readjudication at week 26. As noted above, at the
time these subjects were selected for readjudication, it was obvious
that readjudication of their week 26 remission status would have no
effect on the week 52 sustained remission results for that study arm.
b. The Study Results are Uninterpretable and Cannot Be Salvaged with
Further Analyses
The applicant's improper actions forever taint the study results
reported in the NDA and render them uninterpretable. Further analyses
of the ADVOCATE study cannot salvage this drug approval. If not
controlled and prespecified, the risk of a false positive increases
with every subsequent analysis, making the results less reliable. The
ADVOCATE study's statistical analysis plan did not plan for repeat
analyses of the primary endpoint. This is expected because it is
improper to change the data after an unfavorable analysis and reanalyze
the same endpoint. Therefore, the ADVOCATE study results submitted in
the NDA cannot be considered statistically significant.
Under the ADVOCATE study's statistical analysis plan, the overall
Type I error rate was controlled at 5% significance level (Ref 8 at
30). The original statistical analysis of the study database after
database lock on November 5, 2019, was not significant, returning a
two-sided p-value of 0.1025, which is above the significance threshold.
The statistical analysis submitted for NDA review, which was performed
after changing the study data to reflect the readjudication of five
avacopan subjects, returned a two-sided p-value of 0.0132, suggesting
that avacopan was superior to the control arm and the result was
statistically significant. But as explained below, it was not.
The p-value the applicant reported in the NDA for avacopan was not
controlled for Type I error, but the applicant never disclosed that.
The applicant conducted a second analysis of the primary endpoint after
it changed the data and without controlling for the risk of a false
positive result. The p-value reported in the NDA is therefore
meaningless and uninterpretable.
There is no way to redeem the ADVOCATE study by permitting the
applicant to conduct a new, independent, blinded review and
adjudication of the study data. As stated previously, each re-analysis,
without procedures to control multiplicity that are specified before
the data are locked, will increase the risk of a false positive. This
is exactly what guidelines and regulations around the conduct of
clinical trials are designed to prevent--continuously reanalyzing data
until you get a positive result. Moreover, different adjudications,
even based on the same data, may produce different results because of
the selection and training of the adjudicators, among other things. In
the present case, there may also be biases associated with knowledge of
the study results from prior analyses. With enough readjudication, the
study may inappropriately appear to have statistical significance even
if the drug is not effective. Again, the only analysis that should be
evaluated as part of the TAVNEOS NDA is the original analysis following
database lock on November 5, 2019, which did not demonstrate
statistical significance.
2. Untrue Statements in the Application
The applicant withheld and mischaracterized information about when
and how often the study database was locked, unblinded, analyzed, and
altered. These distortions violated various FDA requirements regarding
the content of NDAs and constituted untrue statements of material fact
that affected FDA's decision to approve the application.
For example, FDA regulations require, as a general matter, that an
NDA contain reports of all investigations of the drug product and ``all
other information about the drug pertinent to an evaluation of the
NDA'' (21 CFR 314.50). These regulations specifically require, among
other relevant data and information: (1) ``the results of statistical
analyses of the clinical trials'' (Sec. 314.50(c)(2)(viii)); (2) a
``description and analysis of each controlled clinical study, and the
documentation and supporting statistical analyses used in evaluating
the controlled clinical studies'' (Sec. 314.50(d)(6)) ; and (3) a
``description and analysis of any other data or information relevant to
an evaluation of the safety and effectiveness of the drug product''
(Sec. 314.50(d)(5)(iv)). Furthermore, the NDA ``is required to contain
a summary of the NDA in enough detail that the reader may gain a good
general understanding of the data and information in the NDA, including
an understanding of the quantitative aspects of the data'' (Sec.
314.50(c)(1)).
Despite these obligations, the applicant did not disclose in the
NDA that the database was locked, unblinded, and statistically analyzed
more than once, nor did the applicant describe or otherwise note the
original, failed primary endpoint analysis. Instead, the applicant
repeatedly and falsely conveyed that there was a single analysis of the
ADVOCATE study, and that it achieved a certain p-value indicating that
the results were statistically significant. In response to an
information request from FDA after

[[Page 23287]]

the NDA was submitted, the applicant then falsely characterized the
study database lock as a single event (``the freeze of the study
database'') occurring on November 20, 2019. For example, as described
in section I.D, the applicant's response to CDER's December 9, 2020,
information request asserts that the ADVOCATE study database was frozen
(i.e. locked) on November 20, 2019 (Ref. 16 at 2, 3). Similarly, the
CSR states that database lock occurred on November 20, 2019 (Ref. 7 at
4). The CSR further states that ``[t]reatment assignments for
individual subjects remained blinded to the study team, Investigators,
and subjects until after the study database had been cleaned and
locked'' (Ref. 7 at 34). That statement is only true if the database
lock referenced is the November 5, 2019, database lock; it is patently
false if the database lock referenced is the November 20, 2019, lock--
the only database lock referenced in the NDA.
Given the material omissions and untrue statements in the
application, the applicant has falsely certified that the data and
information in its application submissions for NDA 214487 were ``true
and accurate.'' The applicant's omissions and untrue statements were
material to the review of NDA 214487, including the determination of
substantial evidence of effectiveness, the overall benefit-risk
assessment, the recommendation of the Director of the Division of
Rheumatology & Transplant Medicine to approve the NDA, and the decision
of the Director of the Office of Immunology & Inflammation to approve
the NDA.

F. Withdrawal of Approval is Appropriate

In light of the facts outlined above, CDER can no longer conclude
that there is, or has ever been, a valid demonstration of substantial
evidence of effectiveness for TAVNEOS. The post hoc nature of the
database changes and involvement of unblinded study personnel, combined
with the resulting change in statistical significance, irrevocably
compromises the credibility of the study results and thus the
demonstration of substantial evidence of effectiveness that supported
the original approval.
The original analysis based on the November 5, 2019, database lock
was not statistically significant. Additional readjudication or
reanalysis of the same data, blinded or unblinded, cannot overcome the
finding of the original analysis. Every additional readjudication or
reanalysis reduces the reliability of the result because of the
increased risk of chance findings. Thus, substantial evidence of
effectiveness of TAVNEOS can no longer be demonstrated with the
existing data.
Furthermore, we are increasingly concerned about the safety profile
of TAVNEOS. In a review of postmarketing data from the applicant's
global safety database, the medical literature, and CDER's Office of
Surveillance and Epidemiology review of FDA's Adverse Event Reporting
System (FAERS) through October 9, 2024, CDER identified 76 cases of
DILI, of which 72 cases were possibly, and 4 cases were probably
causally associated with, avacopan use. Importantly, 7 of the 76 cases
of DILI reported biopsy-confirmed VBDS, a serious and unexpected
adverse event for avacopan. Seventy-four cases reported a serious
outcome, including hospitalizations (n=54) and deaths (n=8). Without
certain demonstration of benefit to balance the serious risks
associated with avacopan, including the risk of hepatotoxicity, the
drug's benefits cannot outweigh its known risks for the conditions of
use for which it is approved.
Accordingly, CDER proposes to withdraw approval of TAVNEOS
(avacopan) capsule, 10 mg, because, on the basis of new information,
evaluated together with the evidence available to FDA when the NDA was
approved, there is a lack of substantial evidence of effectiveness that
the drug will have the effect it is purported or represented to have
under the conditions of use prescribed, recommended, or suggested in
its labeling. This absence of demonstrated benefit is further
aggravated by the drug's known serious risks of hepatoxicity. In
addition, CDER proposes to withdraw approval of the drug because the
application contained untrue statements of material fact, as described
in Section I.E.
We are mindful that TAVNEOS is indicated as an adjunctive treatment
of a serious rare disease associated with high morbidity and mortality.
We also recognize that there are limited therapeutic options for the
condition, and that those options (glucocorticoids) are often not well
tolerated. On the other hand, there are serious safety risks associated
with avacopan, including the risk of hepatotoxicity, with no
demonstration of benefit.
Withdrawing a drug under these circumstances is critical to protect
the public health and the integrity of the drug approval process.
Otherwise, applicants would be incentivized to deceive FDA and, if the
deception is discovered by FDA after approval, claim the drug should
remain on the market while the applicant gets another chance to
reanalyze data and hope for a positive result.

II. Notice of Opportunity for a Hearing

In accordance with Sec. 314.200 (21 CFR 314.200), notice is hereby
given to the applicant and to all other interested persons that the
Director of CDER proposes to issue an order, under sections 505(e)(3)
and 505(e)(5) of the FD&C Act, and 21 CFR 314.150(a), withdrawing
approval of NDA 214487 and all amendments and supplements thereto.
In accordance with section 505 of the FD&C Act and 21 CFR 314.200,
the applicant is hereby provided an opportunity for a hearing to show
why the approval of the NDA should not be withdrawn.
If the applicant decides to seek a hearing, it must file the
following: (1) a written notice of participation and request for a
hearing (see DATES and ADDRESSES) and (2) the data, information, and
analyses relied on to demonstrate that there is a genuine and
substantial issue of fact that requires a hearing (see DATES and
ADDRESSES). Any other interested person may also submit comments on
this notice. The procedures and requirements governing this notice of
opportunity for a hearing, notice of participation and request for a
hearing, the information and analyses to justify a hearing, other
comments, and a grant or denial of a hearing are contained in Sec.
314.200 and in 21 CFR part 12.
The failure of the applicant to file a timely written notice of
participation and request for a hearing, as required by Sec. 314.200,
constitutes an election by the applicant not to avail itself of the
opportunity for a hearing concerning CDER's proposal to withdraw
approval of the NDA and constitutes a waiver of any contentions
concerning the legal status of the drug product. FDA will then withdraw
approval of the NDA, and the drug product may not thereafter be
lawfully introduced or delivered for introduction into interstate
commerce. Any new drug product introduced or delivered for introduction
into interstate commerce without an approved NDA is subject to
regulatory action at any time.
A request for a hearing may not rest upon mere allegations or
denials but must present specific facts showing that there is a genuine
and substantial issue of fact that requires a hearing. If it
conclusively appears from the face of the data, information, and
factual analyses in the request that there is no genuine and
substantial issue of material fact, or if a request for a hearing is
not made in the required format or with the required analyses, the

[[Page 23288]]

Commissioner of Food and Drugs will enter summary judgment against the
person who requests the hearing, making findings and conclusions, and
denying a hearing.
If you choose to submit confidential information under this notice
of opportunity for a hearing, it must be a paper submission filed in
two copies. Except for data and information prohibited from public
disclosure under 21 U.S.C. 331(j) or 18 U.S.C. 1905, the submissions
may be seen at the Dockets Management Staff (see ADDRESSES) between 9
a.m. and 4 p.m., Monday through Friday, and will be posted to the
docket at <a href="https://www.regulations.gov">https://www.regulations.gov</a>.
This notice is issued under section 505(e) of the FD&C Act, 21 CFR
part 314, and under authority delegated to the Director of CDER by the
Commissioner of Food and Drugs.

III. References

The following references marked with an asterisk (*) are on display
at the Dockets Management Staff (see ADDRESSES \1\) and are available
for viewing by interested persons between 9 a.m. and 4 p.m., Monday
through Friday; they are also available electronically at <a href="https://www.regulations.gov">https://www.regulations.gov</a>. References without asterisks are not on public
display at <a href="https://www.regulations.gov">https://www.regulations.gov</a> because they have copyright
restriction. Some may be available at the website address, if listed.
References without asterisks are available for viewing only at the
Dockets Management Staff. Although FDA verified the website addresses
in this document, please note that websites are subject to change over
time.

1. Jayne, D., P. Merkel, T. Schall, et al., 2021, ``Avacopan for the
Treatment of ANCA-Associated Vasculitis,'' NEJM, 384(7): 599-609,
available at <a href="https://pubmed.ncbi.nlm.nih.gov/33596356/">https://pubmed.ncbi.nlm.nih.gov/33596356/</a> (accessed
Apr. 24, 2026).
2. * FDA, ``Arthritis Advisory Committee; Notice of Meeting;
Establishment of a Public Docket; Request for Comments,'' 86 FR
16227 (Mar. 26, 2021), available at <a href="https://www.federalregister.gov/documents/2021/03/26/2021-06265/arthritis-advisory-committee-notice-of-meeting-establishment-of-a-public-docket-request-for-comments">https://www.federalregister.gov/documents/2021/03/26/2021-06265/arthritis-advisory-committee-notice-of-meeting-establishment-of-a-public-docket-request-for-comments</a>
(accessed Apr. 24, 2026).
3. * Expert Report of Marc Walton, MD Ph.D., Sept. 25, 2024, Homyk
v. ChemoCentryx, Inc., Case No. 4:21-cv-03343 (N.D. Ca).
4. * Amgen, Inc., ``Response to FDA Requests Dated Jul. 23, 2025.''
NDA 214487, Aug. 22, 2025.
5. * ChemoCentryx, Inc., ``Protocol'' CL010_168, Amendment 4.0, Jan.
18, 2016, available at <a href="https://cdn.clinicaltrials.gov/large-docs/27/NCT02994927/Prot_000.pdf">https://cdn.clinicaltrials.gov/large-docs/27/NCT02994927/Prot_000.pdf</a> (accessed Apr. 24. 2026).
6. * ChemoCentryx, Inc., ``Statistical Analysis Plan,'' version 2.0,
Oct. 28, 2019, available at <a href="https://cdn.clinicaltrials.gov/large-docs/27/NCT02994927/SAP_001.pdf">https://cdn.clinicaltrials.gov/large-docs/27/NCT02994927/SAP_001.pdf</a> (accessed Apr. 24, 2026).
7. * ChemoCentryx, Inc., ``Clinical Study Report CL010_168.'' NDA
214487, Jun. 1, 2020.
8. * ChemoCentryx, Inc., ``BVAS and VDI Adjudication Committee
Charter,'' Protocol CL010_168, version 4.0, NDA 214487, Jun. 21,
2019.
9. * FDA, Memorandum of Meeting Minutes (IND 120784), Nov. 1, 2016.
10. Love, S., V. Yorke-Edwards, et al., 2021, ``Making a Distinction
Between Data Cleaning and Central Monitoring in Clinical Trials,''
Clin Trials, 18(3): 386-388, available at <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC8174009/">https://pmc.ncbi.nlm.nih.gov/articles/PMC8174009/</a> (accessed Apr. 24, 2026).
11. * FDA, ICH guidance for industry ``E6(R3) Good Clinical
Practice,'' (September 2025), available at <a href="https://www.fda.gov/media/169090/download">https://www.fda.gov/media/169090/download</a> (accessed Apr. 24, 2026).
12. * FDA, International Council for Harmonisation (ICH) guidance
for industry ``E9 Statistical Principles for Clinical Trials,''
(September 1998), available at <a href="https://www.fda.gov/media/71336/download">https://www.fda.gov/media/71336/download</a> (accessed Apr. 24, 2026).
13. * ChemoCentryx, Inc., Study CL010_168, Week 52 Primary Analysis
Topline Tables, Table 14.2.2.1 (Nov. 7, 2019).
14. * ChemoCentryx, Inc., ``Clinical: Response to Information
Request #4.'' NDA 214487, Sept. 25, 2020.
15. * FDA, TAVNEOS (avacopan). ``Information Request 11.'' NDA
214487, Dec. 9, 2020.
16. * ChemoCentryx, Inc., ``Response to Information Request #11:
Clinical & Statistics.'' NDA 214487, Dec. 29, 2020.
17. * FDA, Arthritis Advisory Committee Meeting, Transcript, May 6,
2021, available at <a href="https://web.archive.org/web/20230330025701/https://www.fda.gov/advisory-committees/advisory-committee-calendar/may-6-2021-meeting-arthritis-advisory-committee-meeting-announcement-05062021#event-materials">https://web.archive.org/web/20230330025701/https://www.fda.gov/advisory-committees/advisory-committee-calendar/may-6-2021-meeting-arthritis-advisory-committee-meeting-announcement-05062021#event-materials</a> (accessed Feb. 24, 2026).
18. * FDA, TAVNEOS (avacopan). ``Multi-Discipline Review.'' NDA
214487, Oct. 12, 2018, available at <a href="https://www.accessdata.fda.gov/drugsatfda_docs/nda/2021/214487Orig1s000MultidisciplineR.pdf">https://www.accessdata.fda.gov/drugsatfda_docs/nda/2021/214487Orig1s000MultidisciplineR.pdf</a>
(accessed Apr. 24. 2016).
19. * FDA, ``Approval Letter--TAVNEOS (avacopan) (NDA 214487).''
Letter to ChemoCentryx, Inc., Oct. 7, 2021, available at <a href="https://www.accessdata.fda.gov/drugsatfda_docs/appletter/2021/214487Orig1s000ltr.pdf">https://www.accessdata.fda.gov/drugsatfda_docs/appletter/2021/214487Orig1s000ltr.pdf</a> (accessed Apr. 24, 2026).
20. FDA, ICH guideline for industry ``Structure and Content of
Clinical Study Reports,'' (July 1996), available at <a href="https://www.fda.gov/media/71271/download">https://www.fda.gov/media/71271/download</a> (accessed Apr. 24, 2026).

Tracy Beth H[oslash]eg,
Acting Director, Center for Drug Evaluation and Research.
[FR Doc. 2026-08455 Filed 4-29-26; 8:45 am]
BILLING CODE 4164-01-P

</pre><script data-cfasync="false" src="/cdn-cgi/scripts/5c5dd728/cloudflare-static/email-decode.min.js"></script></body>
</html>

View on FederalRegister.gov →Plain-text source

Indexed from Federal Register on April 30, 2026.

This is legal information, not legal advice. Laws vary by jurisdiction and change frequently. Always verify current law with official sources and consult a licensed attorney in your jurisdiction for advice on your specific situation.