Anti–PD-1 Inhibitor–Related Pneumonitis in Non–Small Cell Lung Cancer

Immune checkpoint blockade with PD-1 inhibitors has revolutionized the treatment of an increasing number of tumor types, including melanoma and non–small cell lung cancer (NSCLC; refs. 1–7). Nivolumab has demonstrated a survival benefit over docetaxel in both squamous (8) and nonsquamous (9) NSCLCs, and was granted FDA approval for squamous NSCLC in March 2015 and for nonsquamous NSCLC in October 2015. Another PD-1 inhibitor, pembrolizumab, has also shown marked antitumor activity in previously treated NSCLC (10) and was granted accelerated FDA approval for PD-L1⁺ NSCLCs in October 2015.

With more widespread prescribing of PD-1 inhibitors, prompt recognition of serious toxicities is necessary for the safe use of these agents. Among immune-related adverse events (irAE) noted during trials of PD-1 inhibitors, pneumonitis has been recognized as an “event of special interest,” occurring at a rate of 3% (9/296) and resulting in three treatment-related deaths (2 patients with NSCLC and 1 patient with colorectal cancer) in a phase I trial of nivolumab (5). The long-term safety in the NSCLC cohort from this phase I trial was updated, and pneumonitis was reported in 7% (9/129), with three pneumonitis-associated deaths (1). In a phase II trial of nivolumab in squamous NSCLC, pneumonitis was one of the most common irAEs, occurring in 5% of patients (6/117), including 4 patients with grade 3 pneumonitis (3).

In response to the increasing awareness of pneumonitis as a serious irAE, our group has described clinical and radiographic features of anti–PD-1 pneumonitis in melanoma patients treated in trials of nivolumab (11). However, this entity has not been previously reported specifically in the NSCLC population. Given the large number of advanced lung cancer patients diagnosed in the United States every year who could potentially be treated with immune checkpoint blockade, and the fact that many symptoms of PD-1 inhibitor–related pneumonitis overlap with common symptoms of lung cancer, clinical and radiographic descriptions of this potentially life-threatening, but treatable, entity are urgently needed.

We report on two cases of anti–PD-1 pneumonitis in advanced NSCLC patients treated with nivolumab after its FDA approval. Improving our understanding of PD-1 inhibitor–related pneumonitis will enable radiologists and oncologists to accurately recognize this entity and promptly provide appropriate treatment.

Among the advanced NSCLC patients treated with nivolumab after its FDA approval as a part of clinical care at our institution, two cases of anti–PD-1–related pneumonitis were identified based on the review of the medical records. The imaging studies of these patients were retrospectively reviewed with an Institutional Review Board–approved clinical research protocol. Chest CT scans at baseline, during therapy, and at follow-up were reviewed by a consensus of three radiologists with expertise in thoracic and oncologic imaging (M. Nishino, N.H. Ramaiya, and H. Hatabu) for findings of pneumonitis, as described (11, 12). CT findings of pneumonitis were assessed for (i) extent in upper, middle, and lower lungs (none, <5%, 5%–25%, 25%–50%, >50%); (ii) distributions in terms of (a) peripheral, diffuse, central, or mixed and (b) upper, lower, diffuse, multifocal, or focal; (iii) lobar involvement; and (iv) specific CT findings, including traction bronchiectasis, consolidation, reticular opacities, ground-glass opacities (GGO), centrilobular nodularity, and honeycombing. In each case, radiographic patterns of pneumonitis were classified according to American Thoracic Society/European Respiratory Society international multidisciplinary classification of interstitial pneumonias and the related conditions as (i) usual interstitial pneumonia pattern; (ii) nonspecific interstitial pneumonia (NSIP) pattern; (iii) cryptogenic organizing pneumonia (COP) pattern; (iv) acute interstitial pneumonia (AIP)/acute respiratory distress syndrome (ARDS) pattern; (v) hypersensitivity pneumonitis pattern; and (vi) not applicable, as described (11–13). Clinical presentation and treatment course of pneumonitis were obtained from the medical record review.

Case 1 is a 72-year-old man with stage IV squamous NSCLC who had disease progression after four cycles of carboplatin and paclitaxel. The patient was then treated with second-line nivolumab monotherapy at a standard dose of 3 mg/kg every 2 weeks. After four doses of nivolumab, the patient reported only mild fatigue but no respiratory symptoms. Chest CT scan at 8 weeks of therapy demonstrated new GGOs, reticular opacities, and consolidation in lower lobes, with a peripheral and lower distribution, representing a COP pattern based on the radiographic appearance (Fig. 1A and B). On the same CT, the dominant right upper lobe tumor had markedly decreased in size compared with baseline, indicating response to therapy.

Over the subsequent weeks while on continued nivolumab therapy, the patient developed progressive dyspnea with cough and wheezing, but no fever. A follow-up chest CT at 15 weeks of therapy showed a significant increase in the radiographic findings which involved all lung lobes (Fig. 1C and D). The distribution of the findings was peripheral and multifocal, representing a COP pattern, and the progressive nature was also indicative of ARDS. Nivolumab was held, and the patient was treated with 60 mg of prednisone by mouth daily as an outpatient, with rapid clinical improvement. He was also placed on trimethoprim-sulfamethoxazole for pneumocystis prophylaxis. Four weeks later, while the patient was on prednisone taper, a repeat scan of the chest showed a marked decrease of consolidations (Fig. 1E and F), and residual GGOs showed a “reversed halo” sign, with central GGO surrounded by dense consolidation of crescentic shape (Fig. 1F; ref. 14). The patient had been on a corticosteroid taper for a total of 8 weeks before prednisone was discontinued altogether, with near-complete resolution of his pneumonitis-related symptoms at that time.

However, within 4 weeks after discontinuation of prednisone, without restarting nivolumab or initiation of any other systemic cancer therapy, the patient again developed respiratory symptoms, including dyspnea, cough, and wheezing, but no fever. Chest CT scan showed bilateral dense consolidations with GGOs and reticular opacities in peripheral and multifocal distributions, again representing a COP pattern (Fig. 1G and H) as noted at the onset of PD-1 pneumonitis. Given the similarities in clinical and radiographic manifestations with the first episode, the findings were most indicative of recurrent pneumonitis after discontinuation of prednisone, or a “pneumonitis flare” in the absence of nivolumab retreatment. The patient restarted prednisone at a dose of 60 mg daily with levofloxacin at a dose of 750 mg by mouth daily, again with improvement in cough, wheezing, and dyspnea. A follow-up CT at 2 weeks on the second course of prednisone treatment showed resolving consolidations and GGOs (Fig. 1I and J).

Case 2 is an 83-year-old woman who originally had a stage IIA NSCLC with neuroendocrine features. She underwent right upper lobectomy and had declined adjuvant chemotherapy. She was subsequently found to have liver metastases and was felt to be a poor candidate for cytotoxic chemotherapy. She was treated with nivolumab, and after 4 weeks of therapy, the patient was admitted to the hospital with an increasing dry cough, dyspnea, and hypoxemia. A CT scan on admission showed new multifocal areas of GGO, reticular opacities, and consolidation in the left lung with a peripheral distribution (Fig. 2A–D), demonstrating a COP pattern based on the radiographic appearance. The liver metastasis decreased in size, in response to nivolumab. She was treated with a prednisone taper, starting at a dose of 60 mg daily, along with sulfamethoxazole-trimethoprim. A chest CT scan taken 2 weeks later showed marked decrease of the multifocal lung opacities. The patient has completed a prednisone taper over the past several weeks. Due to the severity of this patient's pneumonitis, nivolumab was not restarted and she was started on another systemic therapy.

The present report provides the first detailed description of the clinical and radiographic characteristics of PD-1 inhibitor–related pneumonitis in NSCLC patients treated with nivolumab, which was administered as part of standard clinical care and not in the trial setting. The onset of pneumonitis in both cases was rather early, developing within 8 weeks after treatment initiation. Both patients were symptomatic with cough and shortness of breath, and 1 patient experienced hypoxemia. Imaging findings were similar in the two cases, with GGOs and consolidations in a peripheral distribution, which is radiographically indicative of a COP pattern based on the radiographic appearance on CT. Both cases were successfully treated with oral corticosteroids, with marked improvement of clinical symptoms and radiographic findings; however, 1 patient experienced a “pneumonitis flare” after completing a prednisone taper, which was responsive to reinitiation of corticosteroids.

Multifocal consolidations were the major radiographic findings of pneumonitis in the two cases; the consolidations on CT were extensive and developed rapidly within 2 months, which may be a notable feature of the entity. Because of this feature, CT scans of both cases at the onset of pneumonitis were initially interpreted as “infectious pneumonia versus tumor progression.” Although these differential diagnoses are reasonable, treatment-related pneumonitis should also be strongly considered in patients receiving PD-1 inhibitors presenting with new pulmonary findings on imaging (11). The respiratory symptoms of pneumonitis, including cough, shortness of breath, and hypoxemia, overlap with the symptoms of advanced lung cancer itself, providing additional challenges for early identification of anti–PD-1 pneumonitis specifically among NSCLC patients. The increased awareness of the entity among cancer care providers and the multidisciplinary approach, including oncologists, radiologists, and pulmonologists, are necessary for prompt diagnosis and management of this potentially serious adverse event.

There has been one previously published report of the radiographic appearance of anti–PD-1 pneumonitis, which included 3 melanoma patients (11). The report included a spectrum of radiographic and clinical manifestations, with 2 patients showing an AIP/ARDS pattern requiring admission to intensive care units, and 1 patient with a NSIP pattern who was treated with oral corticosteroids as an outpatient and was able to restart nivolumab therapy. Although similarities and differences of the radiographic and clinical features of anti–PD-1 pneumonitis across different tumor types need to be further studied in larger cohorts, the present report describes a COP pattern consistently noted in both patients, indicating it may be one of the leading radiographic patterns of this entity among the NSCLC population, which was not described in the previous report of melanoma patients.

Both patients were successfully treated by oral prednisone taper with clinical and radiographic improvement. However, the patient described in case 1 experienced recurrent pneumonitis after discontinuation of prednisone, with a very similar radiographic COP pattern as seen in the first episode, indicating a “pneumonitis flare” after completing a steroid taper. Notably, a pneumonitis flare occurred without restarting nivolumab, which may be a unique feature of immune-related pneumonitis. Although the detailed mechanisms of this unusual phenomenon remain to be investigated, it may be at least in part related to the durable nature of the effect of immune checkpoint inhibitors that have been shown to be associated with durable and persistent response after drug discontinuation (6). This observation, although limited to one case in our initial experience, raises an important question as to when to discontinue corticosteroids in those who have experienced PD-1 pneumonitis. It may also indicate the needs for additional caution when discussing nivolumab retreatment in these patients.

Prolonged use of corticosteroids raises a concern for a possible negative impact on the antitumor efficacy of immune checkpoint inhibitors. In a recent retrospective review of 298 melanoma patients treated with ipilimumab, 254 (85%) experienced an irAE of any grade, and 103 patients (35%) required systemic corticosteroids to treat an irAE; however, overall survival and time to treatment failure were not affected by the occurrence of irAEs or by the need for systemic corticosteroids (15). Additional studies will be needed to determine if these findings will also extend to patients requiring corticosteroids for the treatment of irAEs after PD-1 inhibitor administration.

Neither patient underwent bronchoscopy or biopsy for pneumonitis. The precise role of bronchoscopy and bronchoalveolar lavage (BAL) in the management of PD-1 pneumonitis is currently unclear, but BAL may help identify an underlying pulmonary infection. Without biopsy, both patients in the present series lack histologic assessments of pneumonitis, which is often the case with patients who have known advanced malignancy receiving systemic therapy (12, 16). A COP pattern of pneumonitis in the present cases was characterized by the radiographic appearance of the lung abnormalities of pneumonitis, based on the assessment of extent, distribution, and specific CT findings, as described previously (11, 12).

In conclusion, in our initial experience with anti–PD-1–related pneumonitis in NSCLC, both cases showed a radiographic COP pattern with extensive consolidation in peripheral distribution. One patient experienced a pneumonitis flare after completing a prednisone taper, which is responsive to a second course of corticosteroids. Further investigations in larger cohorts are needed to characterize the full spectrum of this entity. Given the rapidly growing indications for immune checkpoint inhibitors in a variety of tumor types, knowledge of the clinical and radiographic manifestations of drug-related pneumonitis is essential to maximizing the therapeutic benefit of these agents.

M. Nishino is a consultant/advisory board member for Bristol-Myers Squibb. F.S. Hodi is a consultant/advisory board member for Genentech, Merck, and Novartis. M.M. Awad is a consultant/advisory board member for AstraZeneca, Genentech, and Merck. No potential conflicts of interest were disclosed by the other authors.

Conception and design: M. Nishino, H. Hatabu, F.S. Hodi, M.M. Awad

Development of methodology: M. Nishino, H. Hatabu, M.M. Awad

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Nishino, E.S. Chambers, C.R. Chong, S.W. Gray, J.P. Marcoux, F.S. Hodi, M.M. Awad

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Nishino, N.H. Ramaiya, H. Hatabu, F.S. Hodi, M.M. Awad

Writing, review, and/or revision of the manuscript: M. Nishino, N.H. Ramaiya, S.W. Gray, J.P. Marcoux, H. Hatabu, P.A. Jänne, F.S. Hodi, M.M. Awad

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): F.S. Hodi, M.M. Awad

Study supervision: M.M. Awad

M. Nishino, was supported by 1K23CA157631 (NCI).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.