Kavitha, Ravi —
This memo documents a statistical anomaly I have identified in the NX-7 Phase III dataset as submitted to DCGI on 24 April 2025. I am raising this formally and requesting a data reconciliation meeting within 48 hours, before the DCGI review board completes its assessment.
My concern relates specifically to Batch 7B (sample group BNL/NX7-7B/2847/P3, enrolled November–December 2024, Hyderabad and Chennai sites combined, n=191). I have compared the figures in our submitted DCGI filing against the raw outcome data in the NX-7 Phase III Clinical Data Repository (CDR access: NX7-CDR/LAB). The discrepancies are material.
| Metric | CDR Raw (as of 31 Mar 2025) | DCGI Submission (24 Apr 2025) |
|---|---|---|
| Enrolled subjects (Batch 7B) | 191 | 191 |
| Evaluable subjects | 182 | 193 (9 subjects above enrollment) |
| Objective response rate (ORR) | 61.3% (n=112/182) | 74.2% (n=143/193) |
| Grade 3–4 adverse events | 18 cases (9.9% of evaluable) | 11 cases (5.7% of evaluable) |
| Treatment-related discontinuations | 14 | 8 |
| Batch 7B classification in submission | Flagged as statistical outlier in CDR notes, 15 Feb 2025 | Included as primary data, no outlier notation |
To be direct: the submission data includes 9 subjects not present in the CDR at the time of data cut (31 March 2025). Seven adverse event cases from the CDR are absent from the submission. The Batch 7B outlier notation I added to the CDR on 15 February — recommending exclusion pending re-analysis — has been removed in the version I can now access.
I am not in a position to submit a revised DCGI filing unilaterally, and I am not accusing anyone at this stage. I am requesting that the three of us meet to review the Batch 7B CDR history and the submission data side by side. If there is an explanation I am missing, I want to hear it. If there is not, we have a regulatory obligation to notify DCGI before the review board finalises approval.
I have requested this meeting twice verbally, most recently on 29 April. I am now formalising the request in writing under BNL's internal QA protocol (Policy QA-7.3.2: Mandatory escalation of material data discrepancy concerns). Please confirm a time by 3 May.