Oncology · Medical Literature
Accuracy of Index Lymph Node Pathology in Predicting Overall Response to Neoadjuvant Immunotherapy for Clinical Stage III Melanoma: Results From the Prospective NeoACTIVATE Arm C (NCT03554083) Substudy.
Abstract
For patients with clinical stage III melanoma, a potentially surgically curable disease, neoadjuvant immunotherapy improves event-free survival. Substantial pathologic complete response (pCR) rates in the affected nodal basin have raised interest in response-directed care including de-escalation of the extent of operation and adjuvant therapies. This study prospectively assessed the fidelity of index lymph node (ILN) pathology in predicting pathologic response across the entire nodal basin. The ILN substudy was a preplanned prospective investigation within the multicenter NeoACTIVATE Arm C trial testing neoadjuvant doublet immunotherapy with atezolizumab and tiragolumab followed by therapeutic lymph node dissection (TLND). Eligible patients had resectable clinical stage III nodal disease. Patients were required to have a sonographically visible clip placed in the ILN at diagnosis. Surgeons marked the ILN, which was assessed separately in addition to pathologic assessment of the entire nodal basin using International Neoadjuvant Melanoma Consortium criteria. Among 34 enrolled patients, 30 underwent TLND per protocol. Two patients (6.7%) demonstrated discordance between ILN and nodal basin pathology, both with a pCR in the ILN but residual disease in non-ILNs. The false-negative rate of an ILN pCR was 13.3%. Among eight patients with a single involved lymph node at baseline, the pathologic major response rate was 75% and concordance was 100%. In this prospective substudy, ILN pathology did not fully reflect nodal basin status for all the patients. This finding supports caution in de-escalating treatment based solely on ILN response and highlight the need for prospective trials evaluating oncologic outcomes after limited nodal resection.
Clinical relevance
In patients with clinical stage III melanoma receiving neoadjuvant atezolizumab plus tiragolumab, does pathologic assessment of the clipped index lymph node (ILN) accurately predict pathologic response across the entire therapeutic lymph node dissection (TLND) specimen? __SNAPSHOT__ {"studyDesign":"Preplanned prospective substudy embedded within a single-arm, multicenter, Phase II trial (NeoACTIVATE Arm C; NCT03554083)","setting":"Multicenter; Mayo Clinic and affiliated academic institutions, United States","sampleSize":"n=34 enrolled; n=30 underwent TLND per protocol","population":"Adults with resectable, macroscopic (sonographically visible) clinical stage III nodal melanoma; 76.5% had >1 metastatic lymph node at baseline; 73.5% were Stage IIIC; ILN clip placement at diagnosis required for substudy eligibility","intervention":"Four 21-day neoadjuvant cycles of atezolizumab 1200 mg IV (anti-PD-L1) plus tiragolumab 600 mg IV (anti-TIGIT), followed by TLND and eight adjuvant cycles of atezolizumab 1200 mg IV; ILN marked by surgeon and assessed separately from the remainder of the nodal basin","comparator":"Not applicable (single-arm study); ILN pathology was compared against the pathologic assessment of the entire nodal basin within each patient","primaryOutcome":"Fidelity (concordance rate) of ILN pathology in predicting overall pathologic response across the entire resected nodal basin, assessed using International Neoadjuvant Melanoma Consortium criteria","followUp":"Median follow-up 19.9 months from registration (from the parent Arm C trial); substudy-specific survival follow-up not separately reported","fundingSource":"Institutional research funding from Genentech (reported for lead author); full trial funding source not fully reported in available abstract/summary"}
Key findings
Among 30 patients who underwent per-protocol TLND, two (6.7%) demonstrated discordance between ILN pathology and overall nodal basin pathology. In both discordant cases, the ILN demonstrated pathologic complete response (pCR) while residual disease was identified in non-ILNs, yielding a false-negative rate of 13.3% for ILN pCR in predicting basin-wide pCR. In the pre-specified subgroup of patients with a single involved lymph node at baseline (n=8), the major pathologic response (MPR) rate was 75% and ILN-to-basin concordance was 100%. Across the parent Arm C cohort (n=34), MPR (≤10% viable tumor) was observed in 16/34 patients (47.1%). At 12-month follow-up with a median of 19.9 months from registration, event-free survival (EFS) was 72.0% (95% CI 57.9–89.5%); recurrence-free survival (RFS) was 73.3% (95% CI 56.9–94.5%); and distant metastasis-free survival (DMFS) was 86.0% (95% CI 72.2–100%). Among patients achieving MPR, 12-month RFS and DMFS were both 91.7% (95% CI 77.3–100%). Grade ≥3 adverse events at least possibly related to the neoadjuvant regimen occurred in 2/34 patients (5.9%). NNT/NNH for specific outcomes not calculable from available data. Confidence intervals and p-values for the ILN concordance/false-negative rate endpoints were not reported in the available abstract.
Practice implications
This substudy established that ILN pathology after neoadjuvant anti-PD-L1/anti-TIGIT doublet therapy is highly concordant with the overall nodal basin in most patients, but carries a false-negative rate of 13.3% (2/30) when the ILN shows pCR—meaning approximately 1 in 7–8 patients with an apparent ILN-based complete response would have residual disease elsewhere in the nodal basin. This finding directly informs the safety of response-directed surgical de-escalation strategies (such as omitting TLND in ILN-pCR patients), suggesting that relying solely on ILN pathology to guide TLND omission carries meaningful risk of understaging in patients with multi-node involvement. Crucially, perfect concordance was observed in the single-node-positive subgroup (n=8, 100% concordance), suggesting ILN-guided surgery may be safely considered in this selected subpopulation, though the number is far too small to draw definitive conclusions. Findings apply specifically to patients treated with the atezolizumab/tiragolumab doublet and may not generalize to other neoadjuvant regimens (e.g., nivolumab/ipilimumab or relatlimab-based). Patients with high nodal burden (>1 involved node, Stage IIIC) are likely at highest risk for ILN-basin discordance. Because only an abstract and secondary sources were accessible, specific statistical analyses for the ILN concordance endpoint (e.g., exact confidence intervals for false-negative rate, sensitivity/specificity, McNemar's test) were not available for review. Practice-changing adoption of ILN-guided TLND omission requires additional prospective validation in larger cohorts—an unmet need being actively pursued in the referenced OMIT trial (NCT06754904). __LIMITATIONS__ Very small sample size (n=30 evaluable for the primary ILN substudy endpoint), severely limiting statistical power and precision of the false-negative rate estimate,Single-arm design with no randomized comparator; concordance data are descriptive, not hypothesis-tested with formal inferential statistics in the available report,Four patients enrolled did not undergo per-protocol TLND, introducing potential selection bias in the evaluable population,Substudy embedded in a Phase II trial primarily powered for pathologic response and RFS, not for ILN concordance; post-hoc or exploratory analyses are prone to type I error,Generalizability limited to patients receiving atezolizumab/tiragolumab; concordance rates may differ with other checkpoint inhibitor combinations (e.g., nivolumab/ipilimumab),Predominantly Stage IIIC, multi-node-positive population may overrepresent the highest-risk for discordance; results in lower-risk Stage III patients are uncertain,Follow-up duration for survival outcomes (median 19.9 months) is relatively short for a disease with late recurrence potential,Institutional research funding from Genentech (atezolizumab manufacturer) represents a potential conflict of interest,Only abstract-level data were accessible for full critical appraisal; complete methods, patient-level data, and detailed statistical analyses could not be reviewed __BOTTOM_LINE_CATEGORY__ promising_but_preliminary __BOTTOM_LINE__ In patients with clinical Stage III melanoma receiving neoadjuvant atezolizumab/tiragolumab, ILN pathology was concordant with the entire nodal basin in 93.3% of cases; however, a 13.3% false-negative rate for ILN pCR—meaning residual disease present in non-ILNs despite an ILN pCR—cautions against using ILN pathology alone to guide TLND omission in multi-node-positive patients. Perfect concordance in the single-node-positive subgroup (n=8) is hypothesis-generating but underpowered. Prospective validation in larger trials is required before ILN-directed surgical de-escalation can be adopted as standard of care.