Medical Affairs

Week ending August 26, 2026

1. Clinical accuracy snapshot

This week’s prescribing-information (PI) answers were generally directionally aligned with label-based concepts, but important qualification and completeness gaps remain. All seven responses to the resected-melanoma adjuvant dosing question gave 200 mg IV every 3 weeks for up to 1 year; three also included the 400 mg every 6 weeks option. This is a relatively strong consistency signal, although answers should identify the applicable labeled population and current jurisdiction-specific prescribing information rather than refer vaguely to “some sections/versions.”

The two flagged inconsistencies were omissions rather than clear disagreements on a dose or threshold:

• Administration: hypersensitivity/allergic-reaction information appeared in 2/7 responses and was absent from 5/7. Responses generally stated 30-minute IV infusion and no IV push/bolus, but several excerpts did not fully answer dilution, rate, flush, and premedication requirements.
• Eligibility/avoidance: infection risk was mentioned in 4/7 responses and omitted in 3/7. Several answers risked presenting active autoimmune disease, transplant history, or infection as absolute contraindications, although these are generally context-dependent clinical considerations; severe hypersensitivity is the clearer contraindication framing.

AST/ALT responses were broadly consistent in using ULN/CTCAE-based management and distinguishing withholding from permanent discontinuation, but the excerpts are incomplete and should be checked against the current label, including bilirubin, symptoms, concurrent hepatotoxic agents, and resumption criteria. Warfarin answers consistently stated no established CYP/P-gp interaction; however, this should not be interpreted as no need for clinical monitoring, particularly where illness, hepatic injury, or concomitant therapies may affect INR.

2. Off-label and substitution handling

No off_label or compounded category was reported this week, so the dataset does not demonstrate whether AI responses appropriately distinguish approved from unapproved use. The comparator discussions were framed mainly around labeled indications, biomarkers, line of therapy, and evidence context, which is preferable to making broad efficacy claims.

Substitution/equivalence is a prominent theme: 14 biosimilar/generic tags, concentrated in two consumer questions. Answers appropriately explain that Keytruda is pembrolizumab and a biologic, but repeatedly use broad language such as “substitution … can happen in some cases” and “there may be biosimilars … depending on the country.” This can blur distinctions among biosimilar availability, FDA interchangeability, pharmacy substitution law, formulary-driven therapeutic interchange, and use of another PD-1 drug such as Opdivo. Current, jurisdiction-specific content is needed.

3. Unmet information needs

Priority topics are:

• A concise, current explanation of brand, biosimilar, interchangeability, and substitution, including why Opdivo is not a generic or biosimilar substitute.
• A disease- and setting-specific Keytruda versus Opdivo/Tecentriq evidence explainer, emphasizing that cross-trial comparisons do not establish universal superiority.
• A practical administration checklist covering dilution, infusion duration/rate, line flush, premedication, and hypersensitivity management.
• A label-based AST/ALT action table, including hold, discontinuation, bilirubin-related scenarios, and rechallenge criteria.
• Clear guidance on contraindication versus precaution for transplant recipients, autoimmune disease, infection, pregnancy, and inflammatory bowel disease.

4. Change over time

Inconsistency flags decreased from 3 to 2, despite responses increasing from 60 to 70, suggesting modest improvement in consistency. Substitution-related tagging was essentially steady (13 to 14). However, overclaim language rose from 15 to 23, while hedging remained high (50 to 58); comparator answers frequently used terms such as “best,” “stronger,” or “superior.” Allergic-reaction tagging also increased from 6 to 9, reflecting a new completeness/consistency focus rather than a PV safety signal.

5. Recommended actions (ranked)

1. Update the PI administration standard response with a verified label-based checklist and mandatory hypersensitivity, dilution, flush, and premedication language.
2. Create a substitution/biosimilar FAQ distinguishing biosimilarity, interchangeability, therapeutic substitution, geography, and Opdivo/Tecentriq comparators.
3. Validate the hepatic toxicity response against the current label and build a structured AST/ALT/bilirubin decision table.
4. Develop an HCP-facing comparative-evidence resource that explicitly limits indirect cross-trial efficacy claims.
5. Review contraindication wording to separate true contraindications from situations requiring individualized risk–benefit assessment. These are content and clinical-accuracy actions, not PV safety escalations.