dMMR Endometrial Cancer: Dostarlimab or Pembrolizumab, With No Trial Comparing Them
Two separate trials each found a dramatic benefit from adding immunotherapy to chemotherapy in dMMR endometrial cancer, using two different drugs, with no trial ever comparing them directly. The decision here isn't whether to add immunotherapy. It's choosing between two real, approved answers to the same question.
Dolores V., 63, has run the same small bakery on the same corner for twenty-two years, closing only twice in that span — once for her hip replacement, once now, for a hysterectomy that found Stage IVB endometrioid endometrial carcinoma with peritoneal spread discovered at the same surgery meant to address what everyone had assumed was a benign fibroid. Immunohistochemistry on the resected tumor showed loss of MLH1 and PMS2 staining, confirming mismatch-repair-deficient, microsatellite-instability-high disease — a molecular subtype that has, in the space of about two years, gone from a research curiosity to the single most consequential biomarker in first-line treatment selection for this cancer.
Two separate randomized trials established that consequence, and both are genuinely positive rather than one confirming the other: RUBY added dostarlimab to standard carboplatin-paclitaxel and found a dMMR-subgroup hazard ratio around 0.28 for progression or death, a result so large it is difficult to overstate in a disease where first-line chemotherapy alone has historically offered modest benefit. NRG-GY018 tested the same basic strategy with pembrolizumab instead of dostarlimab and found an almost identical dMMR-subgroup hazard ratio near 0.30. No trial has ever compared the two regimens against each other, which leaves Dolores's actual treatment decision resting not on which drug works — both clearly do, dramatically — but on differences the two trials didn't design themselves to test: RUBY's maintenance dostarlimab continues for up to three years, while GY018's maintenance pembrolizumab is capped around two years or fourteen cycles, a real difference in total treatment burden and infusion-center time for a woman still running a business six days a week between appointments.
The MLH1/PMS2 loss on her tumor carries a second implication distinct from today's drug question: the large majority of MLH1-deficient endometrial cancers — close to nine in ten in the largest pooled series — arise from acquired promoter methylation rather than an inherited mutation, so the prior probability sits heavily against Lynch syndrome before any testing is done. Dolores has been referred for germline testing anyway, because that residual minority is exactly what her sisters and adult daughter would need to know about.
Two positive trials, one decision left over
I'd lean toward dostarlimab. The two trials reported together, but RUBY carried overall survival as a primary endpoint and has since reported mature data — a 3-year overall survival of 78% against 46% in the dMMR group, hazard ratio 0.32 — where GY018's survival data is still immature. The hazard ratios — roughly 0.28 versus 0.30 — are close enough that I wouldn't call this decisive, but between two very similar results, I'd rather rely on the one with more follow-up behind it.
I'd actually lean the other way, for a reason that has nothing to do with efficacy — the two trials are close enough on that front that I don't think maturity of follow-up should be the deciding factor. GY018's maintenance runs roughly two years against RUBY's three. For a woman running a bakery six days a week, a full extra year of ongoing infusions is a real cost, not a trivial one.
"More mature follow-up" is a real consideration in isolation, but when two point estimates are this close together, I don't think it should outweigh a concrete difference in total treatment burden that actually affects her day-to-day life.
I think you're both right that this isn't an efficacy question — RUBY and GY018 answered the same question with results too close together to call a winner on hazard ratio alone. That means the honest thing to do is exactly what neither trial was designed to tell us: lay out the duration and logistics difference for Dolores directly, plainly, and let her choose which tradeoff she'd rather live with.
Agreed, after Dolores was walked through both trials' data and the specific duration difference: carboplatin-paclitaxel with pembrolizumab, continuing maintenance pembrolizumab for approximately two years per the NRG-GY018 schedule.
Fully agreed on the process even though the team started from different starting preferences: all three voices converged on treating this as a patient-preference decision once the efficacy data was shown to be indistinguishable between the two regimens, rather than continuing to argue for a clinical superiority neither trial actually demonstrated.