The role of dacarbazine and temozolomide therapy after treatment with immune checkpoint inhibitors in malignant melanoma patients: A case series and meta‐analysis

Dacarbazine (DTIC) and its oral counterpart temozolomide (TMZ) have been the most used agents in advanced malignant melanoma (MM) patients and they are still used routinely. The preferred first line treatment, immune checkpoint inhibitors (CPIs) might shape the tumor and the tumor microenvironment, possibly affecting the response to subsequent therapies. The aim of this study was to investigate the treatment effect of DTIC/TMZ in MM patients after CPI therapy in a consecutive patient cohort and through systematic literature review and meta‐analysis. Thirty‐five patients with advanced MM treated with DTIC/TMZ after previous CPI therapy in three Swedish regions between 2017 and 2021 were recognized and seven case series studies were identified through systematic database review. Pooled data from all 345 patients showed a median real‐world progression‐free survival (rwPFS) of 1.9 months and overall survival (OS) of 6.0 months. Three of these studies were included in a meta‐analysis comparing DTIC/TMZ after CPI treatment, versus no previous immunotherapy, showing no statistically significant differences in rwPFS or OS but higher real‐world response rate to chemotherapy for the prior‐CPI treated group (Odds Ratio: 2.24; 95% Confidence Interval: 1.04–4.86). The current study supports consideration of DTIC/TMZ in later line of treatment in the immunotherapy era.

| 353   RYDÉN et al.   the inclusion criteria of a randomized study, around 50% treated with a PD-1 inhibitor in the first line were alive at five years (Larkin et al., 2019).The corresponding figure for previously untreated patients with advanced BRAF-mutated MM treated with the combination of BRAF and MEK inhibitors was 34% (Robert et al., 2019).
Hence, there is still a need to further develop and optimize later lines of treatment.
Before the era of immunotherapy and targeted therapies, chemotherapy was the cornerstone in the treatment of advanced MM patients (Garbe et al., 2011;Lee et al., 1995;Wilson & Schuchter, 2016).Dacarbazine (DTIC) and temozolomide (TMZ) have been the most frequently used agents (Garbe et al., 2011;Pasquali et al., 2018;Wilson & Schuchter, 2016), despite the fact that no study has shown a significant improvement in overall survival (OS) (Wilson & Schuchter, 2016).These two drugs still constitute the main treatment option in second or later treatment lines for patients with metastatic cutaneous MM with no driver mutation in BRAF gene, and in third or later lines in cases where BRAF-mutation is present (Wilson & Schuchter, 2016).DTIC and TMZ share the same active metabolite 5-aminoimidazole-4-carboxamide (AIC) (The European Agency for the Evaluation of Medicinal Products.Human medicine European public assessment report (EPAR), 2023) through which they exert an alkylating cytotoxic effect (Pasquali et al., 2018; The European Agency for the Evaluation of Medicinal Products.Human medicine European public assessment report (EPAR), 2023; Quirt et al., 2007; The European Agency for the Evaluation of Medicinal Products-Committee for Proprietary Medicinal Products (CPMP).Summary Information on a Referral Opinion Following an Arbitration for DTIC Faulding, 2002).In a majority of studies, similar clinical outcomes between DTIC and TMZ have been reported (Garbe et al., 2011;Hillner et al., 2000;Pasquali et al., 2018;Patel et al., 2011;Quirt et al., 2007) and the two drugs are generally regarded as interchangeable (Wilson & Schuchter, 2016).
In a study by Nardin et al., DTIC has been reported to promote lymphocyte infiltration and stromal remodeling in cutaneous MM metastases, indicating that DTIC not only exerts its anti-tumor effect through cytotoxic mechanisms, but also by modulating the tumor microenvironment (Nardin et al., 2011).Other studies supporting the theory of immune modulation have followed; Changes in T-cell populations have been shown after TMZ treatment, with decreased levels of CD4+ and regulatory T-cells (Iversen, 2013), and increased CD8+ T-cells in blood (Iversen, 2013;Ugurel et al., 2013).The presence of the latter has proven vital for response to DTIC in mouse melanoma models (Hervieu et al., 2013).
Moreover, Hervieu et al. have shown that DTIC not only upregulates MHC class 1 cytotoxic CD8+ T-cells (Hervieu et al., 2013), but also increases NKG2D ligands on the surface of MM cells, leading to activation of NK-cells and secretion of interferongamma (IFNγ), both found to be essential for antitumoral effect of DTIC in melanoma mouse models (Hervieu et al., 2013).IFNγ has in turn been closely linked to clinical response in patients treated with CPIs (Grasso et al., 2021).The role of IFNγ has also been discussed by Urosevic-Maiwald et al., who investigated MM patients treated with DTIC and sorafenib, and found increased levels of INFγ in serum, and upregulated IFN-stimulated immune response genes in tumor biopsies (Urosevic-Maiwald et al., 2015).
The suggested immune-modulatory effects of DTIC and TMZ, in addition to their alkylating mode of action, could explain why these two substances have gained their place in treatment of advanced MM compared with other cytotoxic chemotherapeutics.The question arises whether CPI treatment could induce persisting alterations in the immune system and the tumor microenvironment that affects response to DTIC and TMZ in later lines of treatment.The existing studies evaluating treatment with DTIC and TMZ after CPI therapy are scarce and show conflicting results (Bouchereau et al., 2021;Gaughan & Horton, 2022;Goldinger et al., 2022;Karachaliou et al., 2020;Klee et al., 2022;Mangin et al., 2021;Weber et al., 2015).
Although some of them are based on large screened patient cohorts, the absolute number of patients who have received TMZ or DTIC after a prior treatment with a CPI is relatively low per study.
The aim of the present study was to investigate the effectiveness of DTIC and TMZ in MM patients after CPI therapy in a consecutive patient cohort and summarize the existing evidence through a systematic review and meta-analysis.

Significance
Checkpoint inhibitors have drastically changed the treatment landscape for patients with advanced malignant melanoma.Nonetheless, primary and secondary resistance is common.The former preferred treatments, dacarbazine and its oral counterpart temozolomide, still constitute later therapeutic options.However, sufficient data on the efficacy of dacarbazine and temozolomide when used after previous immunotherapy is lacking.Checkpoint inhibitors might shape the tumor and the tumor microenvironment, possibly affecting the response to subsequent therapies.
Hence, there is a strong need to clarify whether chemotherapy still has a role to play after immunotherapy failure.
Our results support, still in the immunotherapy era, consideration of DTIC/TMZ treatment.

| Study setting and cohort
All patients with metastatic cutaneous MM or MM with unknown primary site treated with TMZ after treatment with CPIs (PD-1 inhibitors with or without CTLA-4 blocker) in three Swedish

| Outcomes and definitions
Patient outcomes were measured as real-world response rate (rwRR), real-world progression-free survival (rwPFS), and OS.rwRR was defined as any response to treatment as stated in the EMRs by the treating physician's assessment after TMZ treatment initiation.
Real world PFS was defined as time from TMZ initiation to the time of real-world progressive disease (rwPD) or death due to any cause.OS was measured from TMZ treatment initiation to the date of death regardless of cause.

| Searching strategy and study selection process for meta-analysis
Two reviewers performed independent structured searches in the PubMed and Web of Science databases using the following algorithm: ("Melanoma"[Mesh]) AND ("Immune Checkpoint Inhibitors" [Majr] OR "Immunotherapy"[Majr]) AND ("TMZ"[Mesh] OR chemotherapy OR" DTIC" [Mesh]).The date of the search was 20th October 2022.
The same search algorithm, without MeSH standardized keywords for each term, was applied in the Web of Science by two independent reviewers.Discrepancies between reviewers were resolved by discussion.
The following PICO framework was used: Population; advanced cutaneous MM or MM of unknown primary site previously treated with CPIs.
Control; other treatment approach or DTIC/TMZ before CPIs, or no control.
Outcome; Objective response rate (as defined in each individual study), PFS (as defined in each individual study), OS.
Eligible studies were those assessing patients with advanced cutaneous MM or MM of unknown primary site, who received either DTIC or TMZ after treatment with a CPI (PD-1-, PDL-1-and/ or CTLA-4-inhibitors) irrespective of the study type, apart from case reports and literature reviews which were excluded.Studies were also excluded if they described patients treated with unspecified chemotherapy after CPI treatment.Studies describing multiple chemotherapy treatment strategies were included if patient-related outcomes were presented separately for DTIC and/or TMZ.

| Data collection and quality assessment for the meta-analysis
The following data were, when available, extracted from eligible studies: Number of patients, median age and range, ECOG-WHO performance status (PS) at treatment initiation with DTIC/TMZ, proportion of BRAF-mutated patients, proportion of patients with elevated LDH, proportion of patients with brain metastases, rwRR, median rwPFS with corresponding 95% confidence interval (CI), median OS with corresponding 95% CI, and median follow-up time.
For outcome measures from eligible studies, we did not perform any alignment of definitions, but we accepted the proposed definition from each study.The quality of the identified studies was assessed by two reviewers (VR and AV) using the National Institutes of Health (NIH) study quality assessment tool for case series studies (National Heart, Lung, and Blood Institute (NHLBI), 2014).

| Statistical analyses/data synthesis
For the present Swedish study cohort, numbers and frequencies for categorical variables and median with range for continuous variables were used as descriptive statistics.For the time-to-event outcomes rwPFS and OS, Kaplan-Meier curves were used for visualization, and median time with corresponding 95% CI was calculated for the whole cohort and for specific patient subgroups of interest.
For data synthesis in the meta-analysis, pooled RR and a corresponding 95% CI was calculated using random-effects model based on the single-arm studies.For the analysis of rwPFS and OS from single-arm studies, Kaplan-Meier curves were reconstructed and pooled median survival time was calculated using individual patient data from the published Kaplan-Meier graphs using the two-stage approach described by Liu et al. (Liu et al., 2021).
For studies that presented data on the comparative effectiveness of for DTIC/TMZ in patients prior treated with CPIs compared to patients without prior CPIs, we used pooled odds ratio (OR) with 95% CI through the DerSimonian and Laird method for rwRR whereas for the comparisons of rwPFS and OS, the pooled analyses were carried out first by transforming the Hazard Ratios (HRs) and their errors into their log counterparts, and then using the inverse variance method for transforming back into the HR scale.If rwPFS or OS data were unavailable for direct extraction from the primary studies, data were extracted according to the method described by Tierney et al (Tierney et al., 2007).
We assessed the magnitude of heterogeneity among the studies using the I 2 statistics.We considered an I 2 value >50% as indicative of substantial heterogeneity and random effects model was applied for pooled analyses.

| Properties of the present Swedish cohort
Thirty-five patients with unresectable stage III or stage IV MM treated with TMZ after a previous treatment with CPI were identified.Patient characteristics are described in Table 1.Age at diagnosis of non-curable disease ranged between 24 and 74 years with a median of 59. Seventeen patients were female (49%) and median Charlson Comorbidity Index was 2 (range: 0-4).BRAF mutation occurred in 40% of melanomas (n = 14), whereas 46% (n = 16) carried no BRAF mutation and 14% (n = 5) had unknown mutation status.

Swedish cohort
Considering rwRR to TMZ, we found eight patients experiencing complete or partial response (rwRR = 23%), three with stable disease (9%), 22 with rwPD (63%), and two patients who were not evaluable (6%).The median rwPFS was 2.0 months (95% CI: 1.4-2.6 months) (Figure 1a) and median OS was 6.0 months (95% CI: 3.8-8.2) TA B L E 1 Patient-and tumor-related characteristics of a consecutive cohort of 35 malignant melanoma patients in three Swedish regions treated with temozolomide after previous treatment with immune checkpoint inhibitors.(Figure 1b).In exploratory subgroup analyses of potential interest, similar results regarding rwRR and rwPFS as in main analysis were observed (Table 2).Two patients experienced a long duration of response to TMZ with 16 and 27 months, respectively.

| Study selection for meta-analysis, and characteristics of eligible studies
A flowchart over the selection process is illustrated in Figure 2.
Searches in the databases Pubmed and Web of Science resulted in 4502 potential studies.After reviewing titles and abstracts, 4482 papers were excluded.Twenty potentially eligible studies were examined further by reading full texts.Among these, 13 studies were excluded because they failed to meet the inclusion criteria.
After the selection process, seven studies were included in the  3.

| Quality assessment
Assessment of study quality based on NIH study quality assessment tool for case series studies can be found in Table S1.Of the seven studies identified, four studies were assessed as having good quality, two fair and one poor quality.

| Pooled results on effectiveness of TMZ/DTIC after CPI treatment
Eight studies, including our study cohort, generated a patient population of 345 who were treated with DTIC or TMZ after CPI therapy.

F I G U R E 2
Flowchart diagram over study selection process for meta-analysis of studies encompassing patients with advanced malignant melanoma having received treatment with dacarbazine or temozolomide with or without prior checkpoint inhibitor therapy.Note: Characteristics of eligible studies in meta-analysis encompassing patients with advanced malignant melanoma having received treatment with dacarbazine or temozolomide with or without prior checkpoint inhibitor therapy.
a A sub-cohort within the study where patients treated with PD-1 inhibitor immediately before chemotherapy were observed.All patients in the sub-cohort are also represented in the main cohort. b The number also includes eight patients who received a combination of dacarbazine and either fotemustine or carboplatin.

| DISCUSS ION
The purpose of this consecutive case series and meta-analysis was to investigate the effects of DTIC/TMZ-treatment after CPI therapy, and put the results into perspective in the novel immunotherapy era.
The relevance of this topic has increased as of late due to recent evidence of immunomodulatory effects of DTIC/TMZ (Hervieu et al., 2013;Iversen, 2013;Nardin et al., 2011;Ugurel et al., 2013;Urosevic-Maiwald et al., 2015) which could potentially be influenced by previous CPI treatment.We found that DTIC/TMZ seem to at least maintain their effectiveness for all three endpoints -in effect rwRR, rwPFS and OS-in the immunotherapy era.A potential increased likelihood of responding to chemotherapy when used after immunotherapy can be implied but the certainty of evidence for this observation is low and should be interpreted with caution.
The rwPFS and OS from DTIC/TMZ after CPI treatment observed in our consecutive Swedish patient cohort pooled together with the other existing studies showed figures in line with historic results from DTIC and TMZ therapy in first line (Patel et al., 2011).
Generally, response duration to chemotherapy and other anticancer treatments tends to shorten for each line of therapy (Park et al., 2015), and thus, one could argue that the rwPFS and OS from DTIC/TMZ after earlier treatment with a CPI were, although short, longer than expected.
In the meta-analysis, it was observed that the number of treatment responses to DTIC/TMZ was higher when used after CPI than after control.The finding implies an increased likelihood of responding to DTIC/TMZ when used after immunotherapy.
However, there was no significant difference in rwPFS and OS, and compared the treatment response to DTIC/TMZ to patients who had received any kind of CPI treatment at any time before chemotherapy.There was no clear benefit seen in the former group as compared to the latter, and the authors concluded that they could not identify any predictive factors for chemotherapy response after CPI treatment (Goldinger et al., 2022).In line with Goldinger et al.
we could not, in our exploratory analyses, find any subgroup that benefitted from treatment with DTIC/TMZ.
In our study, we focused on DTIC and TMZ monotherapy considering that single-agent DTIC remains the chemotherapeutic with most evidence and has usually represented standard of care when evaluating other treatments (Huncharek et al., 2001).The combination of multiple chemotherapeutic agents, for example DTIC or TMZ combined with other drugs, or carboplatin together with paclitaxel, has in larger meta-analyses shown higher toxicity rates (Pasquali et al., 2018), but no improvement in OS compared to monotherapy (Huncharek et al., 2001;Pasquali et al., 2018).
A combination of CPI and chemotherapy is a common treatment strategy in other malignant diseases (Larroquette et al., 2021), and there are some studies suggesting a beneficial effect in MM as well (Hu et al., 2021;Williamson et al., 2023;Wolchok et al., 2011), however also for these combinations at the cost of increased toxicity (Wolchok et al., 2011).An important aspect to consider when deciding on further systemic oncological therapy in this patient population is patients' quality of life.A careful risk-benefit assessment is essential through weighing the toxicity from a treatment against its benefits, in this patient population for which best supportive care is a valid option.
In accordance with our cohort where two patients showed long-term responses with a duration of 16 and 27 months, re-  have been fit enough, to receive treatment with DTIC/TMZ after CPI therapy.Another aspect to take into consideration was that five patients within the cohort were treated with BRAF + MEK inhibitors between treatment with CPI and TMZ, and we cannot rule out a confounding effect on these patients' outcomes.
While our meta-analysis could only identify retrospective case series with a limited number of cases, it still adds value as the largest study to date to investigate the effect of DTIC and TMZ after CPI treatment.We included all available relevant studies describing treatment with DTIC/TMZ after CPI therapy.However, a challenge and potential limitation of the meta-analysis is that there were some substantial differences between the patient cohorts in included studies, with regard to the number and type of previous treatment lines before DTIC/TMZ, the timespan between CPI treatment and chemotherapy, and the inclusion of patients having received BRAF-/ MEK-inhibitors or not.Lastly, the lack of uniform definition on treatment response across studies is a source of between-study heterogeneity with potential impact on the validity of the pooled analyses.
In conclusion, unresectable stage III and stage IV MM is a disease where the therapy options are few, and there is an unmet need for further therapeutic options after immunotherapy failure and BRAF/ MEK-inhibitor failure in BRAF-mutated MM.The current case series and meta-analysis suggest that treatment with DTIC and its oral counterpart TMZ seems to at least maintain its effectiveness in later lines, thus supporting, still in the immunotherapy era, consideration of this treatment strategy.Our results even imply that the rwRR of chemotherapy might be better when used after previous CPI therapy.A large prospective randomized study to investigate treatment sequencing would be preferable, but cannot be conducted for ethical reasons.
Hence, the best approach to study treatment sequencing in this setting is a meta-analysis of current evidence.
Regions (Region Uppsala county, Region Örebro county, Region Södermanland) between January 1, 2017 and December 31, 2021 were identified and included.No patients treated with DTIC were identified.Eligible patients had discontinued CPI treatment due to either disease progression or toxicity.No specific TMZ dose was used as a cutoff for patient inclusion.A dose of 150-200 mg/m 2 daily for the first 5 days of a 28-day treatment cycle is the recommended dose in patients with metastatic MM according to Swedish National Guidelines (Confederation of Regional Cancer Centres in Sweden, 2023).We excluded patients treated with TMZ prior to CPIs and patients with mucosal or ocular melanoma.Through electronic medical records, the following data was extracted from eligible patients: Age, gender, Charlson Comorbidity Index, occurrence of de novo metastatic disease (stage 4 disease at the time of melanoma diagnosis), occurrence of visceral metastases and central nervous system metastases at the time of CPI initiation, BRAFmutation status, dates of CPI treatment start and discontinuation, CPI antibody, CPI line of treatment, best response to CPI, duration of response to CPI, reason for discontinuation of CPI, treatment with BRAF + MEK-inhibitors (before or after CPI), dates of disease progression and death or end-of-follow-up.The study was performed in accordance with the Helsinki Declaration.All patient data was anonymized immediately upon data extraction.Ethical approval from the Swedish Ethical Review Authority (Reference number: 2019-02469 and 2020-06801) was obtained before data extraction.

F
I G U R E 1 Kaplan-Meier curves in a consecutive cohort, encompassing 5 years (2017-2021), of 35 malignant melanoma patients in three Swedish regions treated with temozolomide after immunotherapy.(a).Real-world progression-free survival.(b).Overall survival.
which calls for caution when interpreting the results.Based on the above-mentioned, one could speculate that CPI treatment might induce changes in the immune system or tumor microenvironment that may facilitate the initial response to DTIC/TMZ, but that these changes are short-lived or rapidly overridden by drug resistance mechanisms.Currently, it is neither known how long after treatment discontinuation an immunomodulatory effect from CPI treatment may last, nor whether there is a "therapeutic window" within a certain period of time after CPI discontinuation, during which it is optimal F I G U R E 4 Kaplan-Meier curves reconstructed from five studies in malignant melanoma patients during treatment with dacarbazine or temozolomide after a previous treatment with an immune checkpoint inhibitor.(a) Real-world progression-free survival.(b) Overall survival.F I G U R E 3 Forest plot on pooled real-world response rate among eligible studies investigating response rate of dacarbazine or temozolomide after treatment with checkpoint inhibitors in malignant melanoma patients in real-world setting.Goldinger a = Dacarbazine-trated patients.Goldinger b = Temozolomide treated patients.confidence interval) from an immunologic point of view to initiate DTIC/TMZ treatment.Goldinger et al., who conducted the largest study incorporated in the meta-analysis, investigated the sub cohort of patients who had received a PD-1 inhibitor as last treatment before chemotherapy,

F
Pooled analyses and comparisons of effectiveness in malignant melanoma patients treated with dacarbazine or temozolomide with or without prior checkpoint inhibitor treatment.(a).Real-world response rates.(b).Real-world progression free survival.(c).Overall survival.