When to Stop Immunotherapy: First Response? Max Response? Toxicity?

Theodore Bosworth


December 12, 2017

Checkpoint inhibitors are an established standard of care for metastatic melanoma, but there is no reliable information about how long to treat. New evidence suggests that many or even most patients are treated too long. Of studies that encourage that conclusion, a recent analysis found that participants in a trial who discontinued their checkpoint inhibitor combination due to toxicity achieved the same benefit as those who completed the planned course.

"Stop at first response? Stop at maximum response? Stop at unacceptable toxicity? Today we cannot tell anything to our patients based on good evidence," said Caroline Robert, MD, PhD, head of the Dermatology Unit at the Institut Gustave Roussy in Villejuif, France. "But I personally think that we are treating many patients for too long with anti-PD-1 therapies, even if the optimal time to stop may not be the same for every patient."

Within healthcare systems, there are ongoing efforts in several countries to formulate recommendations for duration of treatment with checkpoint inhibitor therapy, according to Dr Robert, but she cautioned that these are being pursued without the benefit of blinded and controlled studies or other high-quality evidence needed for definitive guidance. Tackling what is becoming an increasingly controversial topic in an educational session at the 2017 annual meeting of the European Society for Medical Oncology (ESMO), Dr Robert suggested that treatment duration is a critical issue in regard to cost and safety.

The Example of Melanoma

"The good news is that we even have this question in metastatic melanoma. Not so long ago, we treated patients [with available therapies] until there was progression or too much toxicity," Dr Robert said. "This is a new problem."

This question of duration of therapy is also relevant to other cancers in which checkpoint inhibitors are employed, but Dr Robert focused her remarks primarily on melanoma. This is not only her area of expertise, but checkpoint inhibitors also were initially tested in melanoma, so more data and longer follow-up are now available relative to other malignancies. There is also an increasing amount of anecdotal information about outcomes in melanoma patients who discontinue therapy after a partial response (PR), as well as some experience with checkpoint inhibitor re-treatment after relapse.

In the published controlled trials, a variety of regimens have been tested with different checkpoint inhibitors or combination therapies over variable treatment periods. In the initial phase 3 ipilimumab trial, for example, therapy was administered for just 3 months.[1] In the phase 3 pembrolizumab trial, treatment was planned for 2 years.[2] In two phase 3 trials with nivolumab, no limit was placed on treatment duration beyond disease progression.[3]

Although complete responses (CR) were achieved with checkpoint inhibitor therapy in each of these trials, it is not clear whether duration of treatment was important to protection from relapse once an objective response was achieved. In particular, there is growing doubt that further treatment provides benefit once patients obtain a CR.

"We have tried to obtain support for a randomized trial for discontinuing therapy," said Dr Robert, who emphasized that this is the only way in which to obtain level 1 evidence to guide clinical practice. "We did not succeed, but I don't give up hope."

Are Patients Being Treated for Too Long?

A pooled data analysis published just 2 weeks before Dr Robert spoke has provided one clue that patients may be treated too long. This analysis found that patients who discontinued their two-drug checkpoint inhibitor regimen for side effects in the induction phase did as well as those who remained on therapy after induction when a single checkpoint inhibitor was offered as a monotherapy.[4]

In all of the pooled studies, the induction phase consisted of 1 mg/kg nivolumab plus 3 mg/kg ipilimumab administered every 3 weeks. After a four-dose induction, patients began nivolumab monotherapy in a dose of 2 mg/kg every 2 weeks.

After a minimum follow-up of 18 months, progression-free survival (PFS) was not significantly different for the 176 patients who discontinued therapy during induction for adverse events relative to the 233 who continued on therapy (8.4 vs 10.8 months, respectively; P = .097). Median overall survival (OS) has not yet been reached in either group but is not significantly different in the follow-up thus far (P = .23). The objective response rate (ORR) was not just similar in the discontinuation group but numerically higher (58.3% vs 50.2%).

Although Dr Robert said that "it is very impressive to see that the response was maintained without drug" in the discontinuation group, she also noted that these data contribute to growing doubts about the benefit of extended treatment. When PFS and OS curves are displayed graphically, "they are very similar."

Follow-up in the phase 1b KEYNOTE-001 trial with pembrolizumab also suggests that extended benefit is achieved after relatively short periods of treatment, according to Dr Robert. She reviewed outcomes in 67 patients who were permitted to discontinue pembrolizumab after a CR and at least 6 months on therapy. Before discontinuing treatment, patients were asked to sign a consent form acknowledging that they had been informed that there was no evidence that retreatment would be effective in the event of relapse.

Of the 67 patients, 61 (91%) remain in CR after a median follow-up that now exceeds 3 years, according to Dr Robert. Of the six patients not in CR, two patients have died from other causes, leaving only four patients (6%) who have so far progressed after achieving a CR. All four were re-treated, but a PR was achieved in only one patient and there were no CRs. Although Dr Robert said that no conclusions about rechallenge can be drawn from this small number of patients, "the good news is that the CR is maintained for a very long period of time."

These data suggest that most patients stay in remission for extended or indefinite periods after achieving a CR, but stopping a checkpoint inhibitor after a PR or stable disease (SD) "may be more tricky," said Dr Robert. To explore this question, Dr Robert reviewed outcomes in 104 responders in KEYNOTE-006 who stopped pembrolizumab after 2 years of treatment. This group included 24 with CR, 68 with PR, and 12 with SD. Although the median follow-up since stopping the checkpoint inhibitor is only 9.7 months, progression has been seen after only 1 CR, 4 PRs, and 2 SDs.

"This is very encouraging; if you look on the PFS curves, 91% of the patients have maintained benefit despite having stopped the drug," Dr Robert reported.

Current Practice

Definitive information on stopping checkpoint inhibitors is important because these drugs carry an enormous expense and they are associated with a significant burden of adverse events. In the absence of objective evidence, Dr Robert acknowledged that the best practice for stopping checkpoint inhibitors remains unknown, but she described current practice at her center.

"In case of discontinuation due to toxicity with anti PD-1 monotherapy or in combination, we do not rechallenge until the patient has signs of progression," Dr Robert said. "In those who have had at least 6 months of treatment and a confirmed CR, and the patient accepts the risk of stopping—because right now we cannot tell patients that there is no risk—we propose stopping therapy."

For patients with PR or SD who have received 2 years of treatment, stopping therapy is also recommended at Dr Robert's center, even though she reiterated that there are no strong data to back this approach.

"We need to work together to develop clearer information, because this is a very important question," Dr Robert said.

Dr Robert reports financial relationships with Amgen, Bristol-Myers Squibb, Merck, Novartis, and Roche.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.