Dissociated Response and Clinical Impact in Patients Treated With Nivolumab Monotherapy

Immune checkpoint inhibitors (ICIs) are effective for previously treated patients with advanced non-small cell lung cancer (NSCLC). However, an unconventional response pattern is sometimes encountered. A dissociated response (DR), characterized by some lesions shrinking and others growing, has been recognized with ICI treatment. In this study, we examined the characteristics and treatment outcomes of DR in previously treated NSCLC patients, receiving nivolumab monotherapy. We conducted a retrospective cohort study of previously treated patients with advanced NSCLC who received nivolumab. We assessed the tumor response of each organ using the Response Evaluation Criteria in Solid Tumors (RECIST) criteria at the rst radiologic evaluation. We investigated treatment outcome and compared overall survival using the Kaplan-Meier Method and log-rank tests. Further, we conducted the same analysis in patients who had previously received chemotherapy or tyrosine kinase inhibitor therapy in our hospital.

between DR and non-DR cases. Patients showing DR had signi cantly longer overall survival than those showing concordant progressive disease (46.9 vs. 8.2 months, p = 0.038). The frequencies of DR in the ICI, chemotherapy, and tyrosine kinase inhibitor-treated cohorts were 5%, 1%, and 4%, respectively. Conclusion DR was uncommon, but this presented a distinctive pattern of nivolumab response. Some patients might bene t from continuing nivolumab therapy and may achieve a longer overall survival.

Background
Lung cancer is the leading cause of cancer-related deaths worldwide [1]. Non-small cell lung cancer (NSCLC) accounts for approximately 80% of all lung cancers, and in most cases is unresectable and metastatic at the time of initial diagnosis [2]. Recently, immune checkpoint inhibitors (ICIs), which inhibit the programmed cell death 1 (PD-1)/programmed cell death ligand 1 (PD-L1) axis, have demonstrated impressive e cacy in patients with advanced NSCLC. Nivolumab (Bristol-Myers Squibb, Lawrenceville, NJ, USA) has been approved as a standard treatment option for previously treated NSCLC after accumulation of data from the CheckMate-017 and CheckMate-057 clinical studies [3,4].
The response patterns of tumors treated with ICIs may differ from those of tumors treated with conventional chemotherapeutic agents or targeted therapies, owing to their unique mechanism of action [5]. First, an initial are-up followed by tumor shrinkage, de ned as pseudoprogression, has been reported [6,7]. Second, rapid tumor progression following ICI administration, de ned as hyperprogression, has also been described [8,9]. Finally, previous reports have recognized the occurrence of a 'mixed tumor response' phenomenon in some patients, whereby some lesions decrease in size and others grow [10,11]. These heterogeneous response patterns of individual lesions in the same patient have been de ned a "dissociated response (DR)" [12]. This atypical response raises the confusion of stopping or continuing ICI treatment. To date, there are limited data available on the incidence of DR, and its clinical signi cance has not been fully understood or investigated [13].
Thus, in this study, we aimed to demonstrate the prevalence and clinical course of patients exhibiting a DR following treatment with nivolumab and to provide insight on how to potentially improve the management of this group of patients.

Patient inclusion criteria
We retrospectively analyzed patients with previously treated advanced NSCLC, who received nivolumab monotherapy in the Kobe City Medical Center General Hospital between April 2016 and September 2018 ( Figure 1). Patients without any measurable lesions or those whose response to treatment was not evaluated were excluded. Patients who reported never having smoked were de ned as non-smokers, those who had smoked within 1 year of diagnosis were categorized as current smokers, and the remaining patients were considered former smokers. All patients were classi ed on the basis of their clinical stage according to the 8th edition TNM classi cation criteria [14]. Overall survival (OS) was de ned as the period from the day of commencement of nivolumab treatment until death from any cause or the end of the follow-up period. The cutoff date for data collection was December 31, 2020. We isolated tumor DNA from various specimens and analyzed the mutational status of the epidermal growth factor receptor (EGFR) gene at exons 18-21 using the peptide nucleic acid-locked nucleic acid polymerase chain reaction clamp method, as described previously [15]. Anaplastic lymphoma kinase (ALK) translocation was assessed by immunohistochemistry or uorescence in situ hybridization breakapart probes, as described previously [16]. PD-L1 tumor proportion score (TPS) was evaluated by the 22C3 assay [17]. This study was approved by the Ethics Review Board or Institutional Review Board of each participating institute (zn190108). Informed consent was not required owing to the retrospective nature of the study. Tumors were assessed according to two categories: RECIST version 1.1 and iRECIST criteria, which were used in trials to evaluate the e cacy of immunotherapeutics [18,19]. A durable clinical bene t was de ned as the duration of complete response (CR)/partial response (PR) (the sum of the longest diameter [SLD] of the target lesions was decreased by at least 30% from the baseline)/stable disease (SD) over 6 months [20].

De nition of DR
The DR was previously described as a concomitant decrease in tumoral elements and an increase in other elements [12,13,21]. Importantly, no standardized de nition of DR exists in the currently available literature. To extract patients with an apparent DR, we de ned the speci c criteria described below. We reviewed baseline radiographic data and the rst computed tomography/magnetic resonance imaging (CT/MRI) results (performed between 6 to 12 weeks after commencement of nivolumab monotherapy), and evaluated measurable lesions per organ and decided its response. Measurable lesions were de ned as lesions measuring over 10 mm in longest diameter or over 15 mm in short axis diameter for lymph nodes. We examined the tumor response by organs by selecting up to two measurable lesions in one organ. We de ned the DR according to the following criteria: 1) patients who have both CR/PR organs and progressive disease (PD, at least a 20% increase and at least 5 mm from the nadir in the SLD of the target lesions) simultaneously; and 2) patients who had all CR/PR organs but with the appearance of new lesions or apparent deterioration of unmeasurable lesions. All patients received CT scans within 30 days before commencing nivolumab treatment. Lesions treated with concomitant radiation therapy were excluded from this analysis. All radiological images were evaluated by three independent specialists (pulmonologist and radiologist). In cases of disagreement, the radiographic data were re-examined until a consensus was reached through central review.

Pathological analysis of the DR site
We collected pathological specimens of progressive lesions in patients with DR and compared these with the primary lesions. Hematoxylin and eosin-stained sections were evaluated by two experienced pathologists.

Comparison analysis of cytotoxic chemotherapy and tyrosine kinase inhibitor therapy
To clarify the pattern of DR incidence across treatments, we retrospectively analyzed patients who were treated with a rst-line cytotoxic chemotherapy agent or rst-line EGFR tyrosine kinase inhibitor (TKI) in the Kobe City Medical Center General Hospital between April 2016 and September 2018. We reviewed patients with a DR using the de nition described above.

Statistical analysis
Continuous variables were analyzed using the Student's t-test. Dichotomous variables were analyzed using Pearson's χ 2 or Fisher's exact test, as appropriate. Kaplan-Meier's method was used to estimate survival outcomes, which were compared using the log-rank test between the groups. A P-value of <0.05 indicated statistical signi cance. We conducted the statistical analyses using JMP 11 software (SAS Institute, Cary, NC, USA).

Patient characteristics and treatment outcome
Of the 114 patients who were previously treated for advanced NSCLC and who received nivolumab therapy, 7 patients were excluded from the study because they did not have RECIST-de ned measurable lesions or they were not adequately evaluated for tumor response (Figure 1). Finally, 107 patients were included in the study; the patients' characteristics are summarized in Table 1. The mean age of included patients was 68.2 years, and most patients were men (72%), had a history of smoking (75%), Eastern Cooperative Oncology Group performance status (ECOG PS) of 0 or 1 (93%), and adenocarcinoma histology (74%). The total objective response by RECIST criteria was 22%; 19% of the patients had SD, and 59% had PD. At the time of analysis, 47 OS events (44%) had occurred. The median OS and estimated 1-year OS rate were 17.4 months and 58%, respectively.
Characteristics and treatment course of cases with DR DR was observed in 5 patients (5%). The comparison between DR and non-DR patients revealed no signi cant differences in the patients' characteristics (Table 1). Treatment course and a representative CT scan of patients with DR are shown in Figure 2.
Patient No. 6 was a 68-year-old man with advanced lung adenocarcinoma (Figure 2A). He started nivolumab as a second-line therapy. After the four cycles of nivolumab treatment, his primary lesion reduced in size (33 mm to 22 mm), although a new metastatic lesion (10 mm) in his brain was detected on the MRI scan; therefore, DR was diagnosed. His RECIST-de ned tumor assessment was PD due to the new lesion and uncon rmed progressive disease (iUPD) by the iRECIST criteria. Nivolumab treatment was continued; thereafter, his tumor remained stable (iUPD by the iRECIST criteria). After 21 cycles of nivolumab, he developed immune-related diarrhea, and nivolumab was permanently interrupted. His tumors remained stable. After 729 days of nivolumab treatment, he experienced iPD by the iRECIST criteria (PD in primary lesion and brain metastasis).
Patient No. 28 was a 53-year-old woman with advanced lung adenocarcinoma. She began nivolumab as a third-line therapy ( Figure 2B). After 5 cycles of nivolumab, her primary lesion reduced in size; however, her axial lymph node was enlarged (minimal diameter was 15 mm), and DR was diagnosed. Her RECISTde ned tumor response was PD, and her iRECIST-de ned tumor assessment was iUPD. Nivolumab treatment was continued for another 6 cycles, until treatment was discontinued because of the onset of immune-related diarrhea. Her tumor remained stable (iUPD by iRECIST criteria). After 234 days of nivolumab treatment, the tumors were radiologically stable, but her serum levels of carcinoembryonic antigen (CEA) and cytokeratin-19-fragment (CYFRA) were elevated. Her attending physician diagnosed the condition as clinical PD, and she started fourth-line chemotherapy.
Patient No. 40 was an 80-year-old man with advanced lung adenocarcinoma ( Figure 2C), and he was given nivolumab as a third-line therapy. After 3 cycles of nivolumab, his primary lesion was enlarged and pleural effusion was increased, although his pleural dissemination was reduced; thus, DR was diagnosed. His overall tumor assessment by RECIST was SD and that by iRECIST was iSD, and nivolumab treatment was continued. After 3 additional cycles of nivolumab, his primary lesion enlarged and was diagnosed as RECIST-de ned PD. Figure 3 and Supplementary Table 1 Figure 1C). Interestingly, this patient had an adenosquamous histology in the primary lesion (Supplementary Figure 1D); thus, a histological temporal heterogeneity was found between the primary and metastatic lesion.

Response patterns and OS
Patients showing DR had signi cantly longer OS than those showing concordant PD (PD without DR) (46.9 vs. 8.2 months, respectively; p = 0.038). In addition, there was no signi cant difference in the OS between patients who showed concordant PR (PR without DR) and those who showed concordant SD (SD without DR) (not reached and 19.3, respectively; p = 0.24 and p = 0.76, respectively). The datas and Kaplan-Meier curves for OS are shown in Figure 4 and Supplementary table 2.
Comparison between ICI, chemotherapy, and TKI treatment There were 150 patients in the chemotherapy group and 92 patients in the TKI group, and the frequencies of DR were 1% (2/150) and 4% (4/92), respectively (Supplementary Figure 2). Increased frequencies of DR were observed in the ICI and TKI groups compared with that in the chemotherapy group; however, the differences were not statistically signi cant. None of the patients in the chemotherapy and TKI cohorts achieved durable clinical bene t.

Discussion
Our study evaluated the occurrence of DR after commencing nivolumab treatment in patients with previously treated NSCLC. We evaluated the patient characteristics, treatment course after DR, and histological ndings and compared the prevalence of DR in the nivolumab treatment cohort with those in the chemotherapy and TKI cohorts. To the best of our knowledge, this is the rst comprehensive report of DR of ICI monotherapy describing the prognostic bene t of DR.
In our study, 5% of patients with NSCLC exhibited DR after starting nivolumab treatment, and this response was observed across all organs. The DR was rst described as a "mixed response" in melanoma patients in a prospective clinical trial of immunotherapy [22]. Moreover, DR was also observed in a retrospective study of NSCLC cases with a frequency of 7.5% [12]. Importantly, DR is a different phenomenon than pseudoprogression. Pseudoprogression is typically de ned as the initial increase of tumor size or the presence of new lesions prior to a decrease in size [6]. In our cohorts, none of our DR cases showed a decrease in size, and all remained SD. Taken together, we speculate that DR is a distinctive response pattern of ICI monotherapy, irrespective of the cancer type or organ involved.
Such response patterns present a particular challenge for patient management, in terms of whether the patient should switch to next-line treatment, continue the original regimen, be followed-up frequently, or commence local therapy. In our cohort, all patients who exhibited DR continued nivolumab treatment, and 40% of patients (Patients No. 6 and No. 28) achieved a durable clinical bene t [20]. Moreover, better survival was observed in patients with DR than in patients with concordant PD. In previous studies, ICI bene ts were observed in 20-50% of patients who exhibited DR [11,12]. We should be aware that DR does not always represent ICI resistance, and switching ICI treatment to another systemic chemotherapy de ned according to RECIST ver1.1 may be an early decision. However, continuing ICI in carefully selected patients whose clinical conditions are stable and who have not experienced severe toxicities is a possible treatment option. Taken together, using the iRECIST guideline, developed to avoid the underestimations of RECIST ver1.1, may be a better strategy. There is need for further investigation into the potential bene ts and risks of DR in patients who continue immunotherapy.
The underlying mechanisms for a DR have not been understood. A possible explanation is that DR results from immune reactions, such as massive immune cell in ltrations or the sarcoid reaction, which have been reported in previous studies [23,24]. A second possible explanation is that the temporal heterogeneity of the tumor, such as the PD-L1 TPS and immune status, may account for the DR [25]. In our cohort, no speci c immune reactions were observed, and one patient exhibited tumor histology heterogeneity between primary and metastatic lesions.
The DR was previously reported as a relatively common phenomenon in acquired resistance following EGFR TKI monotherapy for EGFR mutation positive NSCLC (identi ed in 20% of patients), and the mechanism was speculated to be the genetic or tumoral heterogeneity [26][27][28][29]. The previously reported prevalence is higher than that in our ndings, as the radiologic de nition is not same. Importantly, DR was considered an independent worse predictive factor in the TKI cohort, and we did not observe any clinical bene t in patients who received chemotherapy or TKI treatment in this study. We should recognize that the clinical signi cance of DR occurring during chemotherapy or TKI therapy is different from that occurring during ICI therapy.
Our study may contribute in establishing appropriate management plans for patients who exhibit DR during ICI treatment; however, it has several limitations. First, it was a retrospective study that included a small number of subjects from a single institution, owing to the rarity of DR. Second, all patients received nivolumab after DR, and we have no data on patients who stopped ICIs or commenced local ablative therapy.

Conclusions
DR was an uncommon but distinctive response pattern in nivolumab monotherapy, and a durable bene t was observed in patients exhibiting a DR. The survival bene t in DR cases was better than in concordant PD cases. These data provide scope for improving ICI therapy for NSCLC patients. Thus, in patients presenting DR, continuing nivolumab therapy may be bene cial and may allow patients to achieve a longer OS. Further research is required to elucidate the characteristics, treatment strategies, and underlying mechanisms of DR.

Abbreviations
CR=complete response; DR=dissociated response; ECOG-PS=Eastern Cooperative Oncology Group performance status; EGFR=epidermal growth factor receptor; ICI=immune checkpoint inhibitor; NSCLC=non-small cell lung cancer; OS=overall survival; PD=progressive disease; PD-1=programmed death protein 1; PD-L1=programmed death ligand 1; PR=partial response; RECIST=Response Evaluation Criteria in Solid Tumors; SD=stable disease; SLD=sum of the longest diameter; TPS=tumor proportion score; TKI=tyrosine kinase inhibitor; iUPD=uncon rmed progressive disease Declarations Funding This research did not receive any speci c grant from funding agencies in the public, commercial, or notfor-pro t sectors.
This study was conducted with the approval of the Kobe City Medical Center General Hospital Ethics Committee (No. zn190108).

Consent for publication
Informed consent was not required owing to the retrospective nature of the study.

Availability of data and materials
All datasets on which the conclusions of this paper rely are available on request.

Authors' contributions
Yuki Sato: guarantor of the paper and responsible for the integrity of the work as a whole, from inception to the published article. Takeshi Morimoto data analysis, interpretation, and revision of manuscript. Shigeo Hara: conducted pathological analysis. Kazuma Nagata: study conception and design. Kazutaka Hosoya, Atsushi Nakagawa, Ryo Tachikawa, and Keisuke Tomii: data acquisition and radiological assessment. All authors have read and approved the manuscript as submitted.