First generation monoclonal antibodies against programmed death receptor 1 (PD-1) and its ligand 1 (PD-L1) have quickly become standard of care in second-line and more recently first-line treatment over chemotherapy for patients with advanced non-small cell lung cancer (NSCLC) and an increasing number of other cancer types [1–6]. Although these drugs are generally better tolerated than chemotherapy, they are associated with unique and variable immune-related adverse events (irAEs) that if not recognized and treated promptly may increase morbidity and rarely cause mortality [7, 8]. We present a case of lung squamous cell cancer (SCC) with complex tumor genomic alterations to highlight the striking yet largely untapped potential of PD-1 inhibitor therapy, as well as the need for further clinical and translational research to optimize the safety and efficacy of cancer immunotherapy.

The patient received docetaxel and an investigational agent on a clinical trial for recurrent disease. Despite an initial partial response after 2 cycles of treatment, the patient had rapid tumor progression radiographically by PET/CT scan after 6 cycles of treatment. He subsequently started on standard of care nivolumab at 3 mg/kg intravenously every 2 weeks based on CheckMate-057 [9]. The patient reported increasing dyspnea on exertion (DOE) and fatigue during his clinic visit for pre-cycle 3 evaluation at week 4 day 3 (i.e., cycle 2 day 10) (Fig. 1A). He denied any productive cough, fever, or night sweats. Chest x-ray on the same day revealed no acute event although it was difficult to compare the complex lung lesions to those on the prior PET and CT scans. The patient proceeded with his third dose of nivolumab at week 5 as planned.

Due to worsening DOE, a follow-up PET/CT scan at week 6.5 was performed (Fig. 1B, b and c compared to a), revealing: (1) interval development of a moderate to large right pleural effusion with adjacent compressive atelectasis, (2) significant increase in the surrounding FDG activity in the preexisting large cavitary lesion in the RLL, and (3) increased RLL consolidation with diffuse FDG activity. It was unclear if the findings in the RLL represented pneumonia, tumor progression, or drug reaction. Interestingly, excluding these lung parenchymal findings, there was significant decrease in the FDG activity and size of known tumors in the bilateral axillary and mediastinal nodes (red arrows in Fig. 1B, b).

The patient was briefly admitted for possible severe pneumonia and received empiric intravenous antibiotics. He was discharged on oral antibiotics after all cultures were negative for infection. However, the patient’s DOE and fatigue did not improve and pneumonitis developed in the LLL of the lung on chest CT scan (Fig. 1B, d). A diagnostic bronchoscopy was performed approximately 4 weeks later, which revealed no evidence of infection and two different pathologic lesions, i.e., organizing pneumonia in the right middle lobe (RML) (Fig. 2a) and extensive endobronchial, moderately differentiated SCC tumors in the RLL bronchus (Fig. 2b), which were debrided from the RLL bronchus with a cryoprobe. Although the patient had received two lines of chemotherapy (gemcitabine and paclitaxel combination and docetaxel) in the prior year with the last dose of chemotherapy that was more than 2 months ago, there was no distinct tumor necrosis or treatment effect noted. There was focal inflammation within the tumor but no distinct tumor cell reaction to the inflammation (Fig. 2a). These findings suggest freshly growing recurrent tumors in the RLL. Conversely, there was no visible tumor seen in the RML and transbronchial biopsy revealed alveolar parenchyma with fibroblast foci, mild collagen expansion of alveolar septa, and non-specific chronic inflammation (Fig. 2b). These features as well as the patient’s clinical course suggested the diagnosis of nivolumab-induced organizing pneumonia. We further determined the PD-L1 expression on these specimens by the FDA-approved IHC assay (PD-L1 22C3 IHC pharmDx, Dako—Agilent Technologies). We found PD-L1-positive cells infiltrating in the RML specimen were predominantly intra-alveolar macrophages (Fig. 2c), which may have contributed to the development of pneumonitis. There was insufficient normal lung tissue present in this specimen for determining if PD-L1 expression on normal lung tissue may also have contributed to the development of site-specific pneumonitis. In contrast, 90% of viable tumor cells in the RLL had high PD-L1 expression, i.e., tumor proportion score (TPS) ≥50% with partial and complete cell membrane staining (Fig. 2d).

Immediately after the initiation of prednisone at 60 mg per day, the patient had symptomatic improvements with decreased DOE and chest tightness and increased energy and exercise tolerance. Chest CT scan about 2 weeks later (i.e., by week 13) revealed dramatic improvement in pneumonitis (Fig. 1B, e). The patient felt that he was nearly back to baseline 3 weeks later. He had a “flare” of pneumonitis symptoms when steroids were quickly tapered off, and subsequently improved with re-initiation of steroids with a slow taper over 2 months. Despite treatment being discontinued after three doses of nivolumab, continued tumor response to a complete remission between week 5 and week 13 (i.e., ~3 months) was evident by radiographic assessment. At the last follow-up before the submission of this report, the patient has remained in clinical remission for 14 months.

The complex diagnostic challenges encountered in this case lead to tumor genomic profiling analysis of four tumor specimens by FoundationOne assay obtained at initial diagnosis, tumor recurrence, and pneumonitis/tumor progression (Fig. 3 and Table 1), a clinical strategy that is not routinely employed for patients with advanced SCC with no driver mutations. We found that all tumor specimens from this patient had a very high tumor mutation burden (TMB) at 87.4–91.0 and 82.9 mut/Mb in the pre- and post-nivolumab tumor specimens, respectively, corresponding to the 95–96 percentile of TMB in lung SCC (Table 1).

The clinical characteristics, radiographic patterns, and treatment course of PD-1 inhibitor-related pneumonitis are quite variable [8, 10, 11]. A recent systematic review and meta-analysis suggests that the overall incidence of pneumonitis during PD-1 inhibitor monotherapy was 2.7% (95%CI, 1.9–3.6%) for all-grade and 0.8% (95%CI, 0.4–1.2%) for grade 3 or higher pneumonitis. The incidence was higher in NSCLC for all-grade (4.1 vs 1.6%; P = 0.002) and grade 3 or higher (1.8 vs 0.2%; P < 0.001) pneumonitis compared to melanoma [11]. The incidence of pneumonitis was more frequent during combination therapy than monotherapy for all-grade (6.6 vs 1.6%; P < 0.001) and grade 3 or higher (1.5 vs 0.2%; P = 0.001) pneumonitis [11]. The median time from therapy initiation to pneumonitis was 2.6 months, although it could occur after one or two doses of a PD-1/PD-L1 inhibitor [11]. Variable clinicopathological factors, such as preexisting lung damage due to tumor burden, smoking, chronic obstructive pulmonary disease, pulmonary fibrosis, and variable expression of PD-L1 on normal lung tissues, have all been associated with a higher incidence of pneumonitis in NSCLC patients compared to other tumor types. However, the implicated factors are highly variable for individual patients [10]. Our patient had grade 1 radiation-induced pneumonitis in the RUL after receiving radiation to the right axilla, which was unlikely to have contributed to the development of grade 3 pneumonitis present in the RLL and LLL. We postulated that the infiltration of PD-L1-positive macrophages played a role in the distribution and severity of pneumonitis in the RML and potentially other lung parenchymal areas. Our finding suggests macrophage-mediated inflammatory responses may have contributed to the pathogenesis of pneumonitis in this area. Early recognition and prompt initiation of high dose, systemic corticosteroids and supportive care in our patient resulted in resolution of pneumonitis without compromise of the antitumor effect. It is unlikely that different clones of PD-1+ CD8+ T cells might have mediated antitumor effect and/or irAE (pneumonitis here) [12], although we could not entirely rule out the expression of PD-L1 on normal lung epithelial cells and other immune regulators in the tumor microenvironment (TME) due to the small size of available tissue specimen obtained by transbronchial fine-needle aspiration. Further mechanistic studies on more cases are needed to fully understand the underlying mechanisms and develop clinical guideline for evaluation and management of PD-1/PD-L1 inhibitor-induced pneumonitis.

The patient’s tumor has several molecular characteristics that have been associated with a favorable clinical response to PD-1 inhibitors. First, the majority of post-nivolumab viable tumor cells stained highly positive (TPS ≥50%) for PD-L1 protein expression by IHC (Fig. 2d). Second, all tumor specimens from this patient had a very high TMB (Table 1). TMB was calculated by counting somatic genomic alterations detected per megabase of the coding region target territory of the FoundationOne test (currently 1.11 Mb for targeted exome sequencing of 315 oncogenes and tumor suppressor genes; Foundation Medicine, Cambridge, MA, USA), after filtering to remove known somatic and deleterious mutations as described previously [13, 14]. High TMB (>16 mut/Mb) by FoundationOne has been independently associated with improved PFS and OS in patients with advanced NSCLC who had received atezolizumab and other PD-1 or PD-L1 inhibitors [15, 16]. Although restricting to the FoundationOne capture regions significantly reduced the total number of genomic alterations detected, the whole exome and FoundationOne targeted exome sequencing counts were highly correlated using The Cancer Genome Atlas (TCGA) bladder urothelial carcinoma exome-sequencing database [14]. Except for one mutation (CDC73 G28*), 13 “functional” genomic alterations, which are defined as known or likely oncogenic genomic alterations, detected in the post-treatment, relapsed tumor (column D) were identical to those of the initial (column A) and recurrent metastatic (columns B and C) tumors. Of note, the mutation allele frequency (MAF) of the identified genomic mutations was usually higher in the recurrent (columns B and C) and progressive (column D) tumors than those of the initial tumor (column A) (Fig. 3). We believe the high MAFs were mainly driven by the fraction of tumor present in the biopsy specimens. These data suggest that it was unlikely that a new neoantigen or tumor-associated antigen (TAA) had emerged in the relapsed tumor compared to the primary or recurrent metastatic tumors in this patient. This phenomenon is quite different from oncogene-driven tumors where stepwise, somatic-cell mutations with sequential, subclonal selections occur during cancer progression and the development of acquired resistance to molecularly targeted therapy [17, 18]. Third, all tumors harbored a PIK3CA Q661K mutation and mutations in several DNA damage response genes (Table 1) that have been associated with increased clinical responses to PD-1 or PD-L1 inhibitors [19–21]. Fourth, among all the functional genomic alterations detected, telomerase reverse transcriptase (TERT or hTERT) promoter -146C>T mutation had incrementally increased in MAF% over time (Fig. 3 and Table 1), which could serve as a TAA. Several therapeutic approaches targeting TERT are under development, including immunotherapies utilizing TERT as a TAA, antisense oligonucleotide- or peptide-based therapies, and TERT promoter-directed cytotoxic molecules [22, 23]. We do not know if whole exome sequencing could identify additional TAA candidates. Further exploration and validation of these molecular biomarkers and potential TAA(s) is warranted.

Discontinuation of nivolumab was recommended in our patient due to the presence of grade 3 pneumonitis. Luckily, at the diagnosis of pneumonitis at 4–5 weeks after initiating nivolumab treatment, we observed radiographic responses of existing tumors (Fig. 1), suggesting the rapid activation of presumably PD-1+, tumor-specific, CD8+ T cells. These potent CD8+ T cells were able to eradicate all established and newly formed biopsy-proven tumors in the RLL by ~3 months. Thus, radiographic evaluation alone did not assess the functional status of host immunity against cancer in our patient after cancer immunotherapy. The continued and sustained antitumor response in our patient beyond a year after discontinuing nivolumab challenges the current clinical recommendation of continuing PD-L/PD-L1 treatment for tumor progression for 2 years. Currently, the patient is under radiographic surveillance every 3–4 months as standard of care for patients with metastatic NSCLC. Moving forward, a noninvasive biomarker assay that can evaluate the status of host immunity against tumor should be developed to evaluate or monitor the status of immune function in cancer patients who have responded to PD-1/PD-L1 inhibitor therapy [20].

Our study has several limitations. First, we did not understand the mechanisms of immune evasion in this patient. Second, we could not identify and characterize the effector immune cells, cytokines, and chemokines that are responsible for the potent and durable antitumor effect in this patient. Identifying the specific T cell receptor (TCR)(s) and HLA-restricted TAA(s) using serial tumor and blood specimens obtained before, during, and after discontinuation of nivolumab in this patient would help to understand the T cell responses that are responsible for such an exceptional antitumor response and could help design a personalized cancer vaccine or adoptive T cell therapy to improve the efficacy and safety of cancer immunotherapy should our patient have tumor recurrence. Third, we could not characterize the expression of tumor cells and immune cells in the TME in more than two areas of the lung that were responsible for the variable tumor responses and pneumonitis. Fourth, we did not determine whether or not other genetic factors, such as genetic polymorphisms in cytokine genes and pharmacogenetics, were involved in modulating the immune responses in our patient [33]. As PD-1/PD-L1 inhibitors have been rapidly integrated into standard of care for NSCLC and increasing numbers of other cancer types [34–38], more mechanistic studies are needed to understand the underlying mechanisms of tumor response and resistance. Also needed is the development of biomarkers, not only to predict treatment effects but also potentially fatal toxicities.

We report our multidisciplinary clinical experience and molecular and immune biomarker studies in a patient with refractory, lung SCC who developed grade 3 pneumonitis after three doses of nivolumab monotherapy concurrent with the onset of a potent antitumor response that led to a durable clinical remission. We found that high PD-L1 expression by IHC staining was seen in intra-alveolar macrophages and viable tumor cells in the pneumonitis and recurrent tumor specimens, respectively. The prompt recognition and institution of oral steroids appeared to have prevented serious morbidity and potential mortality from pneumonitis without compromising the remarkable antitumor effect. Despite discontinuation of treatment after only three doses of nivolumab, continued tumor response to a complete remission at ~3 months evident by radiographic assessments has been maintained up to the time of submission of this report at 14-month follow-up. To the best of our knowledge, this is the first report of comprehensive tumor genomic profiling on serial tumor specimens from a lung SCC patient who has had an exceptional clinical response to nivolumab. There are critical needs (1) to delineate the mechanisms underlying exceptional tumor responses and severe tissue-specific, immune-mediated toxicities, (2) to develop noninvasive assay(s) for evaluating the status of antitumor host immunity, and (3) to develop personalized immunotherapy strategies to improve the effectiveness and safety of cancer immunotherapy.