The Effectiveness of Cancer Immune Checkpoint Inhibitor Retreatment and Rechallenge—A Systematic Review

Simple Summary This systematic review gathered recent findings on immune checkpoint inhibitor retreatment or rechallenge in order to overcome primary resistance. The systematic review was performed according to PRISMA and PICO guidelines. In total, 31 articles were included with a total of 812 cancer patients. There were 16 retreatment and 13 rechallenge studies. Fifteen studies reported improvement or maintenance of overall response or disease control rate at the secondary treatment. Interval treatment, primary response to immune checkpoint inhibitors and the cause of cessation from the primary immune checkpoint inhibitor therapy seemed to be promising predictors of secondary response to immune checkpoint inhibitors. Abstract Despite a great success of immunotherapy in cancer treatment, a great number of patients will become resistant. This review summarizes recent reports on immune checkpoint inhibitor retreatment or rechallenge in order to overcome primary resistance. The systematic review was performed according to PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. The search was performed using PubMed, Web of Science and Scopus. In total, 31 articles were included with a total of 812 patients. There were 16 retreatment studies and 13 rechallenge studies. We identified 15 studies in which at least one parameter (overall response rate or disease control rate) improved or was stable at secondary treatment. Interval treatment, primary response to and the cause of cessation for the first immune checkpoint inhibitors seem to be promising predictors of secondary response. However, high heterogeneity of investigated cohorts and lack of reporting guidelines are limiting factors for current in-depth analysis.


Introduction
The wide introduction of immunotherapy in oncology has significantly changed the prognosis and quality of life of selected cancer patients [1]. The three most commonly registered and used drugs are immune checkpoint inhibitors (ICIs) targeting programmed cell death 1 protein (PD-1), programmed death 1 ligand (PD-L1) and cytotoxic T-lymphocyteassociated protein 4 (CTLA-4) [2]. PD-1 and CTLA-4 are receptors located on T cells which may be bound by their upregulated cognate ligands expressed on tumour cell surfaces, thus suppressing the anti-tumoral immune response. ICIs work by the blockage of their interaction that leads to restoration of cytotoxic T lymphocyte function both in the tumour

Search Strategy
The systematic review was performed according to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) protocol ( Figure 2)

Search Strategy
The systematic review was performed according to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) protocol ( Figure 2) [21] and facilitated a PICO-styled (Patients, Interventions, Comparisons, Outcomes) research question (Table  1) [22].

Evidence Acquisition
On the 3rd of February 2023, we performed a search using PubMed, Web of Science and Scopus. Additionally, Cochrane reviews, Google Scholar and references of included articles were checked for adequate studies. We used the following search query: (immune checkpoint inhibitors OR immunotherapy) AND (rechallenge OR retreatment) AND cancer. Initial search returned 1256 results. All articles were independently reviewed by two researchers (AP, BS) and assessed for inclusion and exclusion criteria. After initial screening, 66 articles were chosen for full-text analysis. In total, 31 met inclusion criteria and were further analyzed.

Inclusion and Exclusion Criteria
We included articles reporting the effects of cancer patient rechallenge or retreatment with ICIs, including case reports and case series. Articles were excluded if authors did not report the oncological effect of ICI administration according to RECIST [23], both at the primary administration and when rechallenged or retreated. Moreover, we excluded articles in which the exact number of rechallenged or retreated patients was not stated. Studies published as abstracts, posters or reports from conferences were excluded from the analysis. Studies in languages other than English were excluded.

PICO
We included studies which met predefined PICO eligibility criteria (Table 1).

Evidence Synthesis
The following information was extracted from original publications and included in Table 2: name of the first author, year of the study, number of patients, stage of disease, CNV metastasis, regimens of primary treatment, treatment between primary and secondary ICI administration, time between ICI rechallenge or retreatment, regimens of ICI rechallenge or retreatment, major outcome of the study and type of cancer treated [23]. Based on reported responses, disease control rate (DCR) and objective response rate (ORR) were calculated and summarized in Table 3, with the exception of case report studies. DCR and ORR were defined as the sum of complete response (CR), partial response (PR) and stable disease (SD), and CR and PR, respectively.     Legend: Atez-atezolizumab; CHT-chemotherapy; CI-confidence interval; DOR-duration of response; ECOG-PS-The Eastern Cooperative Oncology Group Performance Status; gp100-glycoprotein 100; HR-hazard ratio; ICI1-initial treatment with immune checkpoint inhibitors; ICI2-retreatment or rechallenge with immune checkpoint inhibitors; Ipi-ipilimumab; MSI-microsatellite instability; N-number of patients; NA-not available; Niv-nivolumab; NLR-neutrophils to lymphocyte ratio; OS-overall survival; OR-odds ratio; ORR-overall response rate; Pem-pembrolizumab; PD-progression disease; PD-1-programmed death receptor 1; PD-L1-programmed death ligand 1; PFS-progression-free survival; PR-partial response; r/r cHL-relapsed or refractory classical Hodgkin lymphoma; SD-stable disease; TPS-tumour proportion score; TT-targeted therapy; +-combination; /-or; *-lack of clear information; # -according to authors.   Legend: CR-complete response; DCR-disease control rate; NE-not-evaluated; ORR-objective response rate; PD-progressive disease; PR-partial response; SD-stable disease; 1-initial treatment; 2-retreatment or rechallenge; * case report study.

Treatment Regimens
Across included studies, patients were treated with various ICI regimens. Specifically, at the primary treatment nivolumab (17 studies) or pembrolizumab (15 studies) monotherapy was the most frequent. Collectively, most patients were retreated and rechallenged with pembrolizumab (14 studies) or nivolumab (10 studies) monotherapy. A detailed description of regimens or groups of regimens used across all studies is presented in Table 2.

Progression-free Survival and Overall Survival
In Table 4, we collected PFS and OS times for retreated and rechallenged cohorts. In all, except one melanoma [29], one NSCLC [44] and one mixed cancers study [51], retreatment or rechallenge times were inferior to the initial treatment. The median PFS and OS rates at initial treatment were 6.95 months (range: 3.7-24.4) and 21.4 months (range: 15.9-39.6), respectively. The same rates at retreatment or rechallenge were 3.14 months (range: 1.6-23.6) and 15.3 months (range: 6.5-30), respectively. Additionally, we tried to establish median rates for subgroups divided by the cause of cessation of the first ICI; however, due to the high heterogeneity between analyzed studies and missing data, we resigned from applying further statistics.   Legend: CI-confidence interval; CR-complete response; Ipi-ipilimumab; m-median; NA-not available; Niv-nivolumab; OS1-overall survival of initial treatment; OS2-overall survival of retreatment or rechallenge; PFS1-progression-free survival of initial treatment; PFS2-progression-free survival of retreatment or rechallenge; TTP-time to progression.

Treatment Toxicity
In Table 5, we summarized the numbers of ICI adverse events (AEs) observed among retreated and rechallenged patients at both treatments. To compare the rate of AEs, we used a ratio of occurred AEs to number of patients. Hence, median AE rates were 0.69 (range: 0.17-1.63) and 0.58 (range: 0.14-1.5) for initial and retreatment or rechallenge, respectively. On the other side, median severe AE (Grade ≥ 3) rates were 0.21 (range: 0-0.67) and 0.16 (range: 0-0.83), respectively. As stated previously, we resigned from applying further statistics due to high heterogeneity and missing data.  Legend: AE1-adverse effects at initial treatment; AE2-adverse effects at retreatment or rechallenge; G-grade; ICI1-initial treatment with immune checkpoint inhibitors; ICI2-retreatment or rechallenge with immune checkpoint inhibitors; Ipi-ipilimumab; N-number of patients; NA-not available; Niv-nivolumab; Pem-pembrolizumab; SAE1--severe adverse effects at initial treatment; SAE2-severe adverse effects at retreatment or rechallenge; AE and SAE defined as number of events, not patients (SAE = grade ≥ 3).

Discussion
Current knowledge regarding ICI retreatment and rechallenge in oncology is rather limited and it is based on the low number of conducted studies and exploratory data analysis. Hence, the optimal duration of ICI therapy, predictors of response, treatment effectiveness and its safety remain unknown [52]. To summarize the current state of knowledge in this field, we present the first systematic review on ICI retreatment and rechallenge that includes studies of various cancer types and an impressive cohort of 812 patients gathered.
Across analysed studies, the most frequently used ICIs were PD-1 inhibitors pembrolizumab and nivolumab, both at the primary and secondary treatment. Moreover, they share similar three-dimensional structures and effector mechanisms; however, pembrolizumab has higher affinity for recombinant human PD-1 than nivolumab [53,54]. The next widely used drug at the secondary treatment following previous PD-1 or PD-L1 therapy, especially in melanoma patients, was a CTLA-4 inhibitor ipilimumab. However, across five melanoma studies, such a therapeutic strategy showed questionable efficacy [28,30]. Additionally, the usage of CTLA-4 inhibitors in the retreatment setting is contrary to the previously established definition consensus [18]. In summary, ICI retreatment and rechallenge guidelines should be updated for melanoma, but also for the other cancer types in which such a therapeutic strategy is getting more prevalent. It is crucial, especially in the context of the introduction of novel immunotherapies (i.e., anti-II ligand lymphocyte activation gene-3 [LAG-3] [55,56], and tebentafusp [57,58]) which are being extensively investigated in clinical trials and showing promising results in melanoma patients.
Furthermore, the general effectiveness of ICI retreatment and rechallenge is still to be confirmed. In this article, we summarized the ORR and DCR of both primary and secondary ICI treatments for each of individual studies and reported them in Table 2. Additionally, we looked at the cause of cessation for the first ICI. However, due to the high heterogeneity and the lack of statistical values in original publications, the potential pan-cancer statistical analysis would be highly biased. As expected, the efficacy measured by ORR and DCR was lower at the secondary treatment in the majority of studies.However, we identified 15 studies in which at least one parameter has improved [26,34,44,47,51], was stable, or dropped but stayed at a relatively high level (ORR or DCR ≥ 50%) [15,24,25,29,32,35,39,42,46,48]. The first ever study which showed improved ORR or DCR was conducted among eight melanoma patients treated with pembrolizumab or nivolumab and followed by ipilimumab [26]. There was no clear information about the interval treatment in this cohort. On the other side, Kan et al. showed improved responses in four melanoma patients treated with nivolumab and followed with pembrolizumab [34]. All four patients received an interval treatment. On the other side, in our case, while recurred with metastatic lesions in CNS, we followed up the patient with the same ICI agent-pembrolizumab. We believed that long-lasting complete response to primary treatment could be a predictor of response to the secondary treatment, which was in line with previous reports [24,25,32]. Further, Kambhampati et al. showed that in relapsed or refractory cHL patients, avelumab resistance can be overcome with nivolumab or pembrolizumab administration [47]. Finally, in the cohort of various cancer types (predominantly melanoma, lung cancer and lymphoma), Simonaggio et al. showed a slightly improved ORR when using anti-PD-1 or anti-PD-L1 inhibitors. Interestingly, in eight out of ten studies in which ORR or DCR was stable or dropped (but ≥50% rate), cancer patients were retreated or rechallenged with the same ICI [15,24,25,32,35,42,46,48]. This observation, which is in accordance with our own experience with pembrolizumab, may support the administration of the same drug or the same group of drugs twice, in the case of successful and well-tolerated primary treatment. However, it is worth to point out that there could be potential biological differences between patients who never achieved a complete remission and have a relatively brief progression-free interval prior to pretreatment, versus patients with lengthy initial complete remission with late relapses.
Nevertheless, it is necessary to characterize crucial factors that may be predictive for ICI retreatment or rechallenge responses.
Recently, a prior or concomitant radiotherapy [59,60], chemotherapy [61,62] or targeted therapy [63] were proposed as promising strategies to improve the efficacy of primary and secondary immunotherapy. In NSCLC patients, Niki et al., Watanabe et al. and Xu et al., showed that responders to second ICI had undergone chemoradiation, chemotherapy or targeted therapy as an interval treatment [37,41,44]. The primary response to ICI seems to be a valuable factor for predicting the retreatment or rechallenge efficacy. Despite one study reporting lack of impact of primary response [30], there were several studies showing that patients who achieved SD or performed better at primary treatment were also better responding to the secondary ICIs [28,32,33,35,36,44,46]. In contrast, Fujita et al. stated that the failure of atezolizumab retreatment might be due to receiving it as a third-or later-line regimen [39], and the effectiveness is more related to PD-L1 expression rather than interval treatment [38]. In addition, Gobbini et al., showed that patients who did not require the interval chemotherapy achieved better responses at rechallenge [17], which was further supported by a melanoma study in which patients who received targeted therapy as an interval treatment had inferior PFS when retreated [30]. The other independently reported predictive factors were discontinuation of primary ICI due to toxicity or clinical decision rather than PD, which is in line with our analysis [17], and a decrease in neutrophil-tolymphocyte ratio after the intermediate treatment in spite of the increase during primary treatment [34]. Lastly, Wakasugi et al. showed that the superior OS was achieved with salvage ICI when compared to chemotherapy or radiotherapy [48]. Despite the promising efficacy of secondary ICI therapy, the treatment toxicity must also be addressed. Among analyzed studies, the prevalence of adverse events was relatively high and equal approximately 69% and 58% at primary and secondary ICI treatment, respectively. However, in the vast majority of cases these were easily manageable, as the frequency of severe adverse events were 21% and 16%, respectively. As such, ICI retreatment and rechallenge seem to be safe strategies for cancer treatment, which is in line with other reports focused on treatment toxicities [64][65][66].

Future Directions
To further investigate potential future directions of ICI retreatment and rechallenge, we looked at ongoing clinical trials reported in the Clinicaltrials.gov registry. Among phase I clinical trials, we found a phase Ib study of synergistic effect of rechallenge with PD-L1 inhibitor after PD-L1 immunotherapy (NCT05325684) and the study of synergistic response to rechallenge with G-CSF after prior anti-PD-1 treatment (NCT05222009). Further, among phase II clinical trials, we found the trial investigating novel compound Zimberelimab plus lenvatamib in advanced cervical cancer in patients who progressed on or after prior ICIs (NCT05824468), the study of Durvalumab in NSCLC patients who continue the treatment or are retreated with it (NCT04078152), the multi-center study of PD-1 inhibitor combined with hypofractionated radiotherapy and GM-CSF with IL-2 in patients with refractory to prior resistance to PD-1/PD-L1 advanced solid tumours (NCT05530200) and the study investigating retreatment with pembrolizumab after fecal microbiota transplantation in prostate cancer. Also, we found two phase III clinical trials on pembrolizumab rechallenge in melanoma with one of the arms for patients previously treated with pembrolizumab (NCT02362594) and a multimodal study investigating several ICIs such as: Pembrolizumab, Nivolumab, Atezolizumab, Durcalumab and Avelumab with an additional rechallenge arm (NCT04637594). Besides ICI therapies, we found several trials addressing retreatment or rechallenge strategies of targeted therapies (NCT01955681, NCT02514681, NCT00824746), PARP inhibitors (NCT05385068), chemotherapy (NCT00257114) or other treatment modalities. Despite the relatively low number of clinical trials, all of these studies will provide more data and should be followed. Nevertheless, there is a need to design new high-quality studies, taking into consideration the potential and challenges related to this treatment modality.

Conclusions
Evidence supporting all of these findings are limited, with multiple studies missing substantial data on the efficacy of primary and secondary treatment with ICIs. Many details of studies are unknown such as number of patients achieving certain RECIST responses at both ICI treatments, detailed course of drug administration, interval treatment, experienced toxicities and follow-up. Therefore, the specified data should be recorded in databases, which will enable more accurate analyses. Furthermore, the heterogeneity of cancer patients selected for original studies, including distinct prior and secondary oncological regimens, various administration schedules and lack of control cohorts or information on cessation from treatments, highly limits the ability of performing a relevant meta-analysis. More efforts should be made to standardize the treatment regimen between selected patients. All mentioned above must be addressed to effectively and safely translate ICIs into clinics in the nearest future. Thus far, the interval of chemoradiation application, primary response to the treatment as well as the cause of cessation to the first ICI are the strongest factors predicting good response. However, they are insufficient; hence, more studies to identify better predictors are needed.