Rechallenging Immunotherapy After Grade 2 Pneumonitis in a Good Responder
A strong response to first-line pembrolizumab was interrupted by grade 2 immune-related pneumonitis that has now fully resolved. The disagreement is whether to restart a drug that was clearly working, given that pneumonitis recurs on rechallenge at an unremarkable rate but is, of all the immune-related toxicities, the one most likely to kill the patient it recurs in.
Denise K., 59, spent the winter training for her first half marathon, a goal she set the week she finished four cycles of pembrolizumab and got the scan showing her lung tumor had shrunk by nearly half. She never got to run it. Two weeks after that scan, a dry cough and creeping breathlessness on stairs she used to take without noticing sent her back in; chest CT showed new bilateral ground-glass opacities in a pattern consistent with organizing pneumonia, graded as grade 2 immune-related pneumonitis given her symptoms and an oxygen saturation that stayed above 90 percent on room air throughout, never requiring supplemental oxygen or hospital admission. A six-week prednisone taper resolved it completely — repeat imaging is clear, pulmonary function testing has returned to her pre-treatment baseline, and her breathing is back to where it was before any of this started.
The question in front of the team now is whether to restart the drug that was clearly working. The rechallenge literature is more reassuring on frequency than it is on consequence, and the two have to be kept apart. Santini and colleagues’ NSCLC rechallenge series found that roughly half of patients restarted after an immune-related adverse event had no further event, and Dolladille and colleagues’ VigiBase pharmacovigilance analysis put recurrence of the same toxicity at about 29 percent overall — with pneumonitis sitting in the middle of that range, not at the top of it, below hepatitis and comparable to colitis. What separates pneumonitis from those two is not how often it comes back but what it costs when it does: it is the immune-related toxicity with the highest fatality rate in patients on PD-1 and PD-L1 agents for lung cancer, and it is the most frequently observed toxicity in NSCLC rechallenge cohorts because it is also the most common first event in them. That asymmetry — ordinary recurrence probability, disproportionate downside — is what makes her situation harder than a generic restart-or-don’t question about a resolved side effect, and it arrives at a moment when her disease is otherwise responding about as well as first-line therapy could reasonably be expected to. She has already asked directly whether she could simply switch to chemotherapy alone instead of returning to either immunotherapy option, a question the team has not yet fully answered for her.
Restarting a drug that was clearly working
Rechallenge her with pembrolizumab. She had an excellent, clearly documented response before this happened, and Santini and colleagues’ own rechallenge series found that most patients who restart immunotherapy after a resolved immune-related adverse event, including pneumonitis, do not recur. Permanently discontinuing a drug producing this kind of response, for an event that has fully resolved both clinically and radiographically, gives up real, demonstrated benefit for a minority-probability risk.
The majority-do-not-recur framing is accurate for immune-related toxicity broadly, and I will not argue with that overall number.
But you are quoting a frequency at me, and frequency is not the number that should decide this. Dolladille’s recurrence figures do not single pneumonitis out — it comes back at roughly the same rate as hepatitis and less often than colitis. What singles it out is the consequence: of every immune-related toxicity we see on PD-1 and PD-L1 agents in lung cancer, pneumonitis is the one that kills people. A recurrent colitis is a hospitalization and a steroid course. A recurrent pneumonitis that opens at grade 3 instead of grade 2 is a different kind of event entirely, and her first episode tells us her lung is a target organ for this drug in a way we cannot untarget. I would hold immunotherapy permanently and continue with chemotherapy alone.
Both of you are reasoning from the right dataset, but from its population average rather than from what her own case specifically showed us. Her pneumonitis resolved completely within six weeks on a standard taper, with no hypoxia requiring oxygen and no hospitalization — a milder course than the median case those recurrence-rate studies include.
That does not erase the higher baseline recurrence risk pneumonitis carries as a class, but it is a real, case-specific data point that argues for a more favorable individual risk than the population number alone implies. If we rechallenge, do it with a structured plan built around that: a lower symptom-reporting threshold than we used the first time, an earlier repeat CT than routine surveillance would call for, and an explicit, pre-agreed point at which any recurrence, however mild, ends immunotherapy permanently rather than triggering a second debate like this one.
Agreed: rechallenge pembrolizumab with a structured monitoring plan — repeat imaging at three weeks rather than the standard interval, explicit low-threshold symptom-reporting instructions, and a pre-committed rule that any recurrence of pneumonitis, at any grade, ends immunotherapy permanently without reopening the question.
Not agreed: the Pulmonologist accepted the monitoring plan as a reasonable mitigation but remained on record preferring permanent discontinuation as the safer choice, agreeing to the rechallenge as the group’s decision rather than as one they were personally persuaded was correct.