Adjuvant immunotherapy has significantly reduced the chance of recurrence. The Keynote-054 trial revealed that adjuvant pembrolizumab led to a 20% improvement in RFS compared with placebo [4]. However, due to the lack of a deep understanding of PD-1 blockade resistance mechanisms, mechanisms of overcoming resistance and markers for monitoring antitumour activity, some of the patients suffer from recurrent disease during adjuvant therapy or after the end of therapy. The treatment selection of patients who relapse during or after adjuvant immunotherapy is unanswered and includes treatment switch at recurrence, response rates and survival benefit. Owing to a lack of available data, there is no solution other than to follow clinical experience or enrol the patient in a clinical trial after adjuvant therapy failure.

In our study, 59% of the recurrences were distant with or without locoregional relapse. It revealed that pembrolizumab is a more efficient adjuvant therapy in decreasing distant metastases than in decreasing locoregional recurrence, which was consistent with data from the KEYNOTE-054 [6] study. Favourable overall survival may be associated with locoregional recurrence only rather than distant metastases.

A preliminary retrospective analysis was performed to study the patterns of recurrence and salvage therapies following adjuvant anti-PD1 therapy. In this study, 70.5% of the enrolled patients had melanoma recurrence during adjuvant anti-PD1 therapy, and 29.5% of the patients recurred following treatment cessation. The median time to relapse from the commencement of adjuvant therapy was longer than that in a retrospective study conducted by Owen et al. (8 months vs. 4.6 months) [7].

The primary considerations for retreatment after adjuvant PD-1 failure include site of relapse, chronology of adjuvant immunotherapy and relapse, BRAF mutation status, performance status, wishes and financial status of the patient.

The patients who underwent resection alone after the first recurrence had a high relapse rate of 80%. These data suggest that surgery alone in the setting of melanoma relapse after adjuvant immunotherapy may not result in long-term disease control and that adjuvant systemic therapy is needed. It should be noted that no relapses were observed in the three patients treated with adjuvant BRAF/MEKi after resection.

A case report revealed that ipilimumab might be an effective treatment choice after adjuvant immunotherapy failure in a patient with stage IIIC malignant melanoma [8]. A reported phase 2 trial of pembrolizumab combined with ipilimumab resulted in an RR of 29% in metastatic melanoma patients who failed PD-1/L1 therapy [9]. However, a multi-institutional historical cohort study suggested that salvage nivolumab plus ipilimumab and ipilimumab monotherapy showed limited efficacy in Japanese populations with advanced melanoma after PD1 resistance. Nevertheless, ipilimumab is not licenced for the treatment of melanoma in China, and ipilimumab cannot be evaluated as a potential promising regimen.

Owen et al. demonstrated that 40% of the patients who recurred after cessation of adjuvant immunotherapy responded to a rechallenge of anti-PD-1 therapy [7], but 12.5% of the patients in the same setting experienced a response in our study. Of note, only five and eight patients were enrolled in this cohort of two studies. In addition, anti-PD1 agent rechallenge may be an available alternative for patients who have ceased adjuvant immunotherapy.

To evaluate the combination of lenvatinib and pembrolizumab in patients with advanced melanoma with confirmed progression on a PD-1/L1 inhibitor, the LEAP-004 study (NCT03776136) was performed. The overall ORR was 21.4% (95% CI 13.9–30.5), the DCR was 65.0%, and the median DOR was 6.3 months. This phase 2 study revealed that lenvatinib plus pembrolizumab had activity in this population. Similarly, pembrolizumab in combination with an oral multitargeted receptor tyrosine kinase inhibitor may also have activity in selected patients, with a response rate of 22.2% in our study.

In a retrospective study, an overall response rate of 82% was observed in melanoma patients treated with BRAF/MEKi after PD1 failure [7]. BRAF/MEKi also led to a favourable response for BRAF-mutated melanoma in a Japanese population who relapsed after PD1 failure. In this series, our data showed a similar trend, in which patients treated with BRAF/MEKi had an ORR of 75% (3/4).

For advanced melanoma patients harbouring c-Kit mutations or amplifications, imatinib mesylate is an alternative therapy with a favourable response rate of 23.3% [10]. In this study, no patient (0/1) responded to imatinib mesylate.

Chemotherapeutic agents were deemed to have immunomodulatory effects, such as enhancement of tumour immunogenicity and disruption of immune-suppressive pathways [11,12,13]. Chemotherapy drugs and anti-PD-1 agents showed synergistic antitumour activity in non-small cell lung cancer (NSCLC), melanoma and other cancers [14, 15]. A retrospective study demonstrated a superior event-free survival (EFS) benefit (median EFS of 7.6 months in combination regimens vs. 3.4 months in monotherapies) and ORR benefit (ORR 59% in combination regimens vs. 15% in monotherapies) of chemotherapy drugs combined with anti-PD-1 agents compared with monotherapy of chemotherapy drugs or immune checkpoint inhibitors.

In a retrospective study conducted by Fudan University Shanghai Cancer Center, an ORR of 40% was reached in advanced melanoma patients treated with a combination of pembrolizumab with temozolomide as first-line therapy, and 55% of them achieved a durable response [16]. We demonstrated that temozolomide plus pembrolizumab can induce clinical responses in patients whose disease progressed after adjuvant anti-PD-1 monotherapy, resulting in a favourable ORR of 50%. The high ORR and short time-to-response in the temozolomide plus pembrolizumab group suggested that this combination can be used to achieve rapid disease debulking. It is worth noting that the 6 patients treated with temozolomide plus pembrolizumab were from Fudan University Shanghai Cancer Center.

In melanoma patients whose diseases progressed during or after adjuvant pembrolizumab, a higher objective response (approximately 26%) to subsequent chemotherapies was observed [15]. In this study, we demonstrated an improved ORR of the combination of carboplatin and paclitaxel compared with that of chemotherapy historic controls in the preimmunotherapy era [17].

Additionally, we revealed that the combination of temozolomide and pembrolizumab after adjuvant anti-PD-1 therapy failure is superior to chemotherapy in the same setting.

From these data, the options for first disease recurrence occurring during adjuvant anti-PD1 monotherapy or after completion of 1-year adjuvant anti-PD1 agents include surgery with adjuvant therapy, BRAF/MEK inhibitors (for BRAF mutant patients), pembrolizumab in combination with an oral multitargeted receptor tyrosine kinase inhibitor, and chemotherapeutic agents alone or with pembrolizumab. Rechallenge with a PD-1 inhibitor is available for patients who relapse after the completion of adjuvant PD-1 inhibition.

Despite including three large cancer centres, we could only recruit a modest number of patients. The follow-up time and imaging methods were not stringently regulated, and a centralised review was not performed. Due to these limitations, our results need to be further validated using a prospective study with a large sample size to minimise the heterogeneity in the population.