Lung Stereotactic Body Radiation Therapy and Concurrent Immunotherapy: A Multicenter Safety and Toxicity Analysis

Abstract

Purpose:

Radical treatment of metastases with stereotactic body radiation therapy (SBRT) is commonly implemented in patients receiving concurrent immune checkpoint inhibition (ICI), despite limited safety and toxicity data. The purpose of this study was to evaluate the safety and tolerability of lung SBRT with concurrent ICI.

Methods and Materials:

Records from a single academic institution were reviewed to identify patients treated with lung SBRT and concurrent (within 30 days) ICI; a contemporaneous cohort receiving lung SBRT alone was included for reference. Treatment-related adverse effects occurring within 30 days (acute) and 180 days (subacute) of SBRT were evaluated.

Results:

Our study included 117 patients; 54 received SBRT with concurrent ICI (56 courses, 69 target lesions), and 63 received SBRT alone (68 courses, 79 lesions). Median follow-up was 9.2 months in the SBRT + ICI cohort. Among the patients, 67.9% received ICI monotherapy, 17.9% ICI/chemotherapy, and 14.3% ICI/ICI combinations; 25% received ICI between SBRT fractions, and 42.9% received ICI both before and after SBRT. The risk of grade 3 pneumonitis was higher in the SBRT + ICI versus SBRT alone cohort (10.7% vs 0%, P < .01) and any-grade pneumonitis was similar (33.9% vs 27.9%, SBRT + ICI vs SBRT, P = .47). The risk of any-grade pneumonitis appeared elevated with ICI/ICI combinations (62.5% vs 29.2%). Receipt of ICI, planning treatment volume, and lobes involved in SBRT were linked to high-grade pneumonitis. Subacute grade 3+ adverse effects occurred in 26.8% of SBRT + ICI and 2.9% of SBRT-alone patients.

Conclusions:

Overall, concurrent lung SBRT + ICI is safe. Given the clinically meaningful risk of pneumonitis, closer monitoring should be considered for SBRT + ICI patients, especially those receiving radiation therapy with ICI/ICI combinations.

Introduction

Radical-intent treatment of limited metastatic disease has demonstrated survival advantages over standard-of-care therapies in multiple randomized studies.^1–4 Accordingly, the use of stereotactic body radiation therapy (SBRT) for 1 or more metastatic deposits has recently seen increased interest and utilization. Immune checkpoint inhibitors (ICI) have extended the survival of patients with advanced malignancies across a wide spectrum of primary tumor types.^5,6 Immune-related adverse effects (AEs) unique to ICI have been described.⁷

SBRT and ICI combinations have been encouraged as a strategy to augment the systemic immune response.^8–10 However, concerns based on overlapping toxicity profiles have limited the widespread use of SBRT administered concurrently with ICI.¹¹ Reports of AEs in the context of thoracic radiation therapy (RT) and ICI have been inconsistent. Some have reported significantly higher rates of treatment-related pulmonary toxicity, whereas others have seen modest to no increase.^12–15 These studies have often examined immunotherapy AEs in the setting of advanced malignancies after palliative-intent, conventionally fractionated RT.

The toxicity profile of lung SBRT and concurrent ICI is not well defined, despite increased use of and interest in this approach. The purpose of this study was to evaluate the safety and tolerability of lung SBRT with concurrent ICI, compare its profile to lung SBRT alone, and identity relevant factors associated with toxicity.

Methods and Materials

Patients who received lung SBRT at 5 centers within a single academic institution were reviewed to identity those who received concurrent immune checkpoint inhibition. The review and collection of patient, disease, and treatment factors and outcome data for patients treated with lung SBRT were performed under an institutional review board—approved protocol. Patients received lung SBRT per institutional standards. Patients were simulated with 4-dimensional computed tomography. The extent of tumor motion was estimated based on 4-dimensional computed tomography, and appropriate motion management strategies were used. A Varian Real-Time Position Management system was used for phase-based gating (Varian Medical Systems, Palo Alto, CA). Patients were immobilized using custom-made SBRT-specific systems, including CIVCO with abdominal compression or abdominal belt or Medical Intelligence setups. Lung SBRT dose, volume, and fractionation were at the discretion of the treating physician based on treatment indication, tumor location size, proximity to adjacent organs at risk, normal tissue tolerance, and receipt of prior thoracic radiation. The conformality index, as defined by the ratio of the prescription isodose volume to the planning treatment volume (PTV), was calculated for each lesion. D2cm, defined as the maximum dose at 2 cm from the PTV in any direction and expressed as percent of the prescription dose, was also extracted. SBRT plan quality was evaluated by both conformality metrics using guidelines as described in the protocol documents for NRG RTOG 0813.

Relevant demographic, disease, and treatment variables, including technical characteristics of radiation delivery, were captured and maintained in a single database. To account for differences in dose and fractionation, biologic effective dose, assuming an α/β of 10 (BED10), was determined for all treatment courses. Concurrent administration of immunotherapy was defined as within 30 days of SBRT. Development of AEs, including pneumonitis, was determined using all available records, including treatment visit notes during SBRT, follow-up chest imaging, and multispecialty follow-up visit with radiation oncology, medical oncology, thoracic surgery, or pulmonology. If applicable, inpatient hospitalizations were noted. AEs were graded using the Common Terminology Criteria for Adverse Events Version 5.0. The earliest date of pneumonitis onset was noted. Development of treatment-related AEs was defined as either acute (within 30 days of SBRT treatment completion) or subacute (within 180 days of the end of SBRT). A contemporaneous cohort of patients who received lung SBRT alone was included as a reference cohort to identify the impact of concurrent ICI.

For each treatment cohort, descriptive statistics were generated for categorical and numerical variables. For each treatment cohort, the rates of pneumonitis and high-grade (3 or higher) pneumonitis were compared across categorical variables using χ² tests or Fisher’s exact tests, where appropriate, and across numeric variables using analysis of variance. Data from both cohorts were combined, and patient characteristics, including pneumonitis, were compared between treatment groups. To determine the relationship between PTV and RT dose, the Pearson correlation coefficient was used in testing PTV against BED10. To examine the relationship between lung dosimetric endpoints and risk of pneumonitis, the mean lung dose, dose received by 1000cc (D1000cc), D500cc, and D250cc were determined. Each nominal dose result was converted to EQD2. Additionally, the volume of the lung receiving the biologically effective doses of 2.5 Gy (V_BED2.5), V_BED5, V_BED10, V_BED15, and V_BED20 were also calculated. A receiver operating characteristic analysis was performed to identify optimal cut points for differentiating any grade pneumonitis for each dosimetric endpoint. Cut points that maximized the combined sensitivity and specificity (geometric mean) were selected. The area under the curve and odds ratios (ORs) with 95% confidence intervals were determined for each dosimetric endpoint. All analyses were performed using SAS 9.4 (Cary, NC); significance was assessed at the .05 level.

Results

Patients were treated with lung SBRT between June 2012 and January 2019. Fifty-six treatment courses, corresponding to 69 target lesions across 54 patients, were identified from those receiving lung SBRT with concurrent immunotherapy. Sixty-eight courses of treatment across 63 patients (79 target lesions) were included, comprising the SBRT-alone cohort. Median follow-up length was 9.23 months in the primary (SBRT + ICI) cohort and 16.67 months in the SBRT-alone cohort. Patients in the SBRT + ICI group were younger (median age 64 vs 69 years, P < .001), more like to have received prior thoracic irradiation (35.7% vs 17.65%), and received a greater number of total radiation courses (23.21% vs 7.35% receiving 2 or more prior courses, P < .001). The SBRT + ICI cohort had a greater proportion of patients with better performance status (85.7% vs 66.2% patients with performance status 0-1). Other characteristics were similar between the 2 cohorts and are summarized in Table 1.

Table 1.

Demographic, disease, and treatment characteristics for SBRT + ICI and SBRT-alone cohorts

Characteristic	SBRT + ICI, n = 56 (%)	SBRT, n = 68 (%)	P value
Age (median), y	64	69	<.001*
Sex			.624
Female	26 (48.15%)	34 (53.97%)
Male	28 (51.85%)	29 (46.03%)
Race			.530
White	35 (62.5%)	43 (63.24%)
Black	18 (32.14%)	24 (35.29%)
Other	3 (5.36%)	1 (1.47%)
Performance status			.074
0	9 (16.07%)	6 (8.82%)
1	39 (69.64%)	39 (57.35%)
2	7 (12.5%)	18 (26.47%)
3	1 (1.79%)	5 (7.35%)
Lesions			.947
1	48 (85.71%)	58 (85.29%)
≥2	8 (14.29%)	10 (14.71%)
Location			.267
LLL	10 (17.86%)	9 (13.24%)
LUL	9 (16.07%)	22 (32.35%)
RLL	14 (25%)	15 (22.06%)
RML	4 (7.14%)	1 (1.47%)
RUL	17 (30.36%)	18 (26.47%)
Media.	2 (3.57%)	3 (4.41%)
Lobes involved			.728
1	53 (94.64%)	63 (92.65%)
2	3 (5.36%)	5 (7.35%)
Centrality			.109
Peripheral	28 (41.18%)	43 (54.43%)
Central	40 (58.82%)	36 (45.57%)
Prior lung RT			.022*
Yes	20 (35.71%)	12 (17.65%)
No	36 (64.29%)	56 (82.35%)
Prior RT courses			<.001*
0	21 (37.5%)	48 (70.59%)
1	22 (39.29%)	15 (22.06%)
≥2	13 (23.21%)	5 (7.35%)
Prior lung SBRT			.877
Yes	6 (8.93%)	6 (8.82%)
No	50 (89.29%)	62 (91.18%)
Lung disease			<.001*
Yes	5 (8.93%)	35 (51.47%)
No	51 (91.07%)	33 (48.53%)
Smoking status			.400
Current	8 (14.81%)	11 (17.46%)
Prior	33 (61.11%)	43 (68.25%)
Never	13 (24.07%)	9 (14.29%)
Dose and fractionation
10 Gy × 5	30 (43.5%)	30 (38%)
12 Gy × 4	6 (8.7%)	33 (41.8%)
18 Gy × 3	3 (4.3%)	10 (12.7%)
6 Gy × 5	12 (17.4%)	2 (2.5%)
10 Gy × 3	3 (4.3%)	0 (0%)
9 Gy × 3	12 (17.4%)	0 (0%)
7.5 Gy × 8	0 (0%)	2 (2.5%)
10 Gy × 2,	0 (0%)	1 (1.3%) each
8 Gy × 3
9 Gy × 5,	1 (1.3%) each	0 (0%)
12 Gy × 3,
8 Gy × 7

Abbreviations: ICI = immune checkpoint inhibition; LLL = left lower lobe; LUL = left upper lobe; Media. = mediastinum; RLL = right lower lobe; RML = right middle lobe; RT = radiation therapy; RUL = right upper lobe; SBRT = stereotactic body radiation therapy.

^*Statistical significance (P < .05).

The majority (90.7%) of patients treated with SBRT + ICI had metastatic disease and received treatment to site(s) of progression. Lung cancer was the most common primary tumor site (18 adenocarcinoma, 10 squamous cell carcinoma, 8 small cell carcinoma, and 3 other). Eight (14.3%) patients received SBRT + ICI to more than 1 target, and 3 (5.36%) received SBRT to lesions involving more than a single lobe. A total of 7 ICI agents were used with lung SBRT; the mostly commonly used was pembrolizumab in 22 patients (39.3%). ICI/ICI and ICI/chemotherapy combinations are detailed in Table 2. A total of 32.2% of patients received ICI combined with another systemic agent, 14.3% in the form of ICI/ICI combinations. A total of 25% of patients received ICI on days between SBRT delivery, and 42.9% of patients received a sandwich approach (both before and after SBRT).

Table 2.

Details of concurrent SBRT + ICI treatment cohort

Characteristic	SBRT + ICI, N = 56 (%)
Indication
Metastatic—progression	49 (90.7%)
Metastatic—consolidation	2 (3.7%)
Recurrent/nonmetastatic	5 (9.3%)
Primary tumor
Lung	38 (70.4%)
Melanoma	8 (14.8%)
Head and neck	3 (5.6%)
Urothelial, renal, adrenal, sarcoma, unknown carcinoma	1 (1.6%) each
Combination therapy
ICI monotherapy	38 (67.9%)
ICI/ICI combination	8 (14.3%)
ICI/chemotherapy combo.	10 (17.9%)
Timing
Interdigitated	14 (25%)
Before and after RT	24 (42.9%)
Before RT alone	2 (3.6%)
After RT alone	16 (28.6%)
ICI agents
Pembrolizumab	22 (39.3%)
Nivolumab	15 (26.8%)
Atezolizumab	8 (14.3%)
Durvalumab	6 (10.7%)
Tremelimumab	5 (8.9%)
Ipilimumab	4 (7.1%)
Cemiplimab	4 (7.1%)
ICI/chemotherapy agents
Pembrolizumab/pemetrexed	4 (40%)
Cemiplimab/cyclophosphamide	4 (40%)
Atezolizumab/bevacizumab	1 (10%)
Pembrolizumab/carboplatin/paclitaxel	1 (10%)
ICI/ICI agents
Durvalumab/tremelimumab	5 (62.5%)
Ipilimumab/nivolumab	3 (37.5%)
Prior lines of systemic therapy
0	5 (8.9%)
1	27 (48.2%)
2	18 (32.1%)
3	4 (7.1%)
4	2 (3.6%)
Prior taxane use
Yes	12 (21.4)
No	44 (78.6)
Prior RT overlap
Yes	10 (17.9)
No	46 (82.1)
Conformality index
No deviation	56 (91.8)
Minor deviation	5 (8.2)
Major deviation	0 (0)
D2cm
No deviation	43 (70.5)
Minor deviation	13 (21.3)
Major deviation	5 (8.2)
PTV, mean (SD), cm3	83.05 (94.75)
GTV, mean (SD), cm3	32.22 (58.38)
Tumor size, mean (SD), cm	3.33 (2.22)

Abbreviations: GTV = gross tumor volume; ICI = immune checkpoint inhibition; PTV = planning target volume; RT = radiation therapy; SBRT = stereotactic body radiation therapy.

As expected, lesions with larger treatment volumes were treated to lower biologic effective doses (Pearson correlation coefficient, −0.361; P = .006). The majority of treatment plans achieved the appropriate conformality indices. When evaluated using conformality index, 91.8% of lesions were sufficiently conformal, meeting the recommended values in the NRG RTOG 0813 protocol, with the remaining 8.2% of lesions having minor deviations. As evaluated using D2cm, 70.5% of lesions met protocol specifications, with 21.3% having minor deviations. Of the 8 patients who received treatment for more than a single target lesion, only a single patient had meaningful dose overlap, as defined by any overlap of the 50% isodose volume from the separate targets; this patient did not develop pneumonitis of any grade.

The incidence of acute AEs, occurring predominantly during SBRT delivery, was similar between the SBRT + ICI and SBRT-alone groups (Table 3). However, SBRT + ICI patients were more likely to have treatment-related AEs up to 90 days after SBRT; grade 3 (G3) or higher subacute toxicity was seen in 26.8% versus 2.9% of patients (P < .001). Severe (G3 or higher) pneumonitis was substantially higher when concurrent ICI was given (10.7% vs 0%, P = .007). Median time to any-grade pneumonitis onset was 3.1 months from end of SBRT + ICI and 2 months for G3 or higher. Median time to pneumonitis onset was 3.4 months for SBRT alone. ICI was discontinued owing to toxicity in 13 of 56 (23.2%) patients’ treatment courses. Of 6 patients who developed G3+ pneumonitis in the SBRT + ICI cohort, treatments received were corticosteroids (in 5 of 6 patients, 83.3%), supplemental oxygen (83.3%), and antibiotics (50%). Five out 6 demonstrated clinical improvement, with median time to improvement of 7 days (range, 2-23). Median time to return to G1 pneumonitis was 36 days, based on completion of the steroid taper.

Table 3.

Toxicity summary of lung SBRT and concurrent ICI

Toxicity	SBRT, N = 68 (%)	SBRT + ICI, N = 56 (%)	P value
Pneumonitis
Any grade	19 (27.9%)	19 (33.9%)	.472
Grade 3 or higher	0 (0%)	6 (10.7%)	.007*
Highest grade acute AE			.217
0	31 (45.6%)	25 (44.6%)
1	29 (42.6%)	28 (50%)
2	8 (11.8%)	2 (3.6%)
3	0 (0%)	1 (1.8%)
Highest grade subacute AE			<.001*
0	19 (27.9%)	5 (8.9%)
1	24 (35.3%)	16 (28.6%)
2	23 (33.8%)	20 (35.7%)
3	2 (2.9%)	12 (21.4%)
4	0 (0%)	3 (5.4%)

Abbreviations: AE = adverse event; ICI = immune checkpoint inhibition; SBRT = stereotactic body radiation therapy.

^*Statistical significance (P < .05).

The risk of any-grade pneumonitis appeared elevated with ICI/ICI combination therapy (62.5% vs 29.17%, P = .105); however, the risk of high-grade pneumonitis was similar compared with other ICI treatment types (12.5% vs 10.42%). ICI administration between SBRT treatment days was safe; it did not appear to increase the risk of high-grade or any-grade pneumonitis over nonoverlapping delivery (Tables 4 and 5). Higher mean sum PTV (121.99 cm³ vs 63.11 cm³, P = .027) was associated with any-grade pneumonitis. Similarly, SBRT involving 2 lung lobes was associated with greater risk of high-grade pneumonitis (66.7% vs 7.55%, P = .028). The risk of G3+ and any-grade pneumonitis was not predicted by ICI agent, prior lung RT, overlapping radiation volumes, lesion centrality, number of SBRT target lesions, prior chemotherapy use, or smoking status. High-grade pneumonitis was associated with treatment involving more than a single lobe, receipt of ICI, and PTV, using the combined cohort of all patients (Table 6). The performance of 9 dosimetric points as potential predictors of pneumonitis was evaluated by receiver operating characteristic testing. There were no statistically significant predictors of pneumonitis (Table E1, available online at https://doi.org/10.1016/j.ijrobp.2019.12.030). Using a threshold EQD2 dose of 13.43 Gy being received by 500 cm³ of the lung, the OR for pneumonitis was 2.65 (95% confidence interval, 0.79-8.87; P = .11).

Table 4.

Predictors of any-grade pneumonitis in SBRT + ICI cohort

Characteristic	No pneumonitis, N = 37 (%)	Pneumonitis, N = 19 (%)	P value
Age (median), y	65	63.5	.645
Sex			.778
Female	18 (64.2)	10 (35.71)
Male	19 (67.86)	9 (32.14)
Race			.891
White	24 (68.57)	11 (31.43)
Black	11 (61.11)	7 (38.89)
Other	2 (66.67)	1 (33.33)
Performance status			.423
0	7 (77.78)	2 (22.22)
1	23 (58.97)	16 (41.03)
2	6 (85.71)	1 (14.29)
Lesions			1
1	32 (66.67)	16 (33.33)
≥2	5 (62.5)	3 (37.5)
Location			.499
LLL	6 (60)	4 (40)
LUL	5 (55.56)	4 (44.44)
RLL	12 (85.71)	2 (14.29)
RML	2 (50)	2 (50)
RUL	11 (64.71)	6 (35.29)
Media.	1 (50%)	1 (50%)
Lobes treated			.263
1	36 (67.92)	17 (32.08)
2	1 (33.33)	2 (66.67)
Centrality			.644
Peripheral	14 (70)	6 (30)
Central	23 (63.89)	13 (36.11)
ICI/ICI combination			.105
No	34 (70.83)	14 (29.17)
Yes	3 (37.5)	5 (62.5)
ICI timing			.338
Interdigitated	26 (61.9)	16 (38.1)
Noninterdigitated	11 (78.57)	3 (21.43)
ICI agent			.556
Atezolizumab	6 (75)	2 (25)
Cemiplimab	2 (50)	2 (50)
Durvalumab	2 (33.33)	4 (66.67)
Ipilimumab	3 (75)	1 (25)
Nivolumab	9 (75)	3 (25)
Pembrolizumab	15 (68.18)	7 (31.82)
Prior lung RT			.293
Yes	15 (75)	5 (25)
No	22 (61.11)	14 (38.89)
Prior RT courses			.559
0	14 (66.67)	7 (33.33)
1	13 (59.09)	9 (40.91)
≥2	10 (76.92)	3 (23.08)
Prior lung SBRT			.208
Yes	31 (62)	19 (38)
No	6 (100)	0 (0)
Prior RT overlap			1
Yes	7 (70)	3 (30)
No	30 (65.2)	16 (34.8)
Prior chemotherapy			.324
Yes	35 (68.6)	16 (31.4)
No	2 (40)	3 (60)
Prior taxane use			.189
Yes	10 (83.3)	2 (16.7)
No	27 (61.4)	17 (38.6)
Lung disease			1
No	34 (66.67)	17 (33.33)
Yes	3 (60)	2 (40)
Smoking history			1
No	13 (92.86)	1 (7.14)
Yes	32 (91.43)	3 (8.57)
Smoking current			1
Yes	37 (90.24)	4 (9.76)
No	8 (100)	0 (0)
PTV sum (mean), cm3	83.48	127.13	.027*

Abbreviations: ICI = immune checkpoint inhibition; LLL = left lower lobe; LUL = left upper lobe; Media. = mediastinum; PTV = planning target volume; RLL = right lower lobe; RML = right middle lobe; RT = radiation therapy; RUL = right upper lobe; SBRT = stereotactic body radiation therapy.

^*Statistical significance (P < .05).

Table 5.

Predictors of high-grade pneumonitis in SBRT + ICI cohort

Characteristic	No high-grade pneumonitis, N = 50 (%)	High-grade pneumonitis, N = 6 (%)	P value
Age (median), y	64	65	.923
Sex			.669
Female	24 (85.71)	4 (14.29)
Male	26 (92.86)	2 (7.14)
Race			.573
White	32 (91.43)	3 (8.57)
Black	15 (83.33)	3 (16.67)
Other	3 (100)	0 (0)
Performance status			.681
0	9 (100)	0 (0)
1	34 (87.18)	5 (12.82)
2	7 (87.5)	1 (12.5)
Lesions			.200
1	44 (91.67)	4 (8.33)
≥2	6 (75)	2 (25)
Location			.629
LLL	10 (100)	0 (0)
LUL	7 (77.78)	2 (22.22)
RLL	13 (92.86)	1 (7.14)
RML	4 (100)	0 (0)
RUL	14 (82.35)	3 (17.65)
Media.	2 (100)	0 (0)
Lobes treated			.028*
1	49 (92.45)	4 (7.55)
2	1 (33.33)	2 (66.67)
Centrality			.405
Peripheral	19 (95)	1 (5)
Central	31 (86.11)	5 (13.89)
ICI/ICI combination			1
No	43 (89.58)	5 (10.42)
Yes	7 (87.5)	1 (12.5)
ICI timing			1
Interdigitated	37 (88.1)	5 (11.9)
Noninterdigitated	13 (92.86)	1 (7.14)
ICI agent			.948
Atezolizumab	7 (87.5)	1 (12.5)
Cemiplimab	4 (100)	0 (0)
Durvalumab	5 (83.33)	1 (16.67)
Ipilimumab	4 (100)	0 (0)
Nivolumab	10 (83.33)	2 (16.67)
Pembrolizumab	20 (90.91)	2 (9.09)
Prior lung RT			1
Yes	18 (90)	2 (10)
No	32 (88.89)	4 (11.11)
Prior RT courses			.216
0	17 (80.95)	4 (19.05)
1	20 (90.91)	2 (9.09)
≥2	13 (100)	0 (0)
Prior lung SBRT			1
Yes	44 (88)	6 (12)
No	6 (100)	0 (0)
Prior RT overlap			1
Yes	9 (90)	1 (10)
No	41 (89.1)	5 (10.9)
Prior chemotherapy			1
Yes	45 (88.2)	6 (11.8)
No	5 (100)	0 (0)
Prior taxane use			1
Yes	11 (91.7)	1 (8.3)	1
No	39 (88.6)	5 (11.4)
Lung disease			1
No	46 (90.2)	5 (9.8)
Yes	4 (80)	1 (20)
Smoking history			1
No	13 (92.86)	1 (7.14)
Yes	37 (88.1)	5 (11.9)
Smoking current			.575
Yes	41 (87.23)	6 (12.77)
No	9 (100)	0 (0)
PTV sum (mean), cm3	76.36	138.76	.129

Abbreviations: ICI = immune checkpoint inhibition; LLL = left lower lobe; LUL = left upper lobe; Media. = mediastinum; PTV = planning target volume; RLL = right lower lobe; RML = right middle lobe; RT = radiation therapy; RUL = right upper lobe; SBRT = stereotactic body radiation therapy.

^*Statistical significance (P < .05).

Table 6.

Predictors of high-grade pneumonitis in combined cohort

Characteristic	No high-grade pneumonitis, N = 118 (%)	High-grade pneumonitis, N = 6 (%)	P value
Age (median), y	67	65	.340
Sex			.682
Female	61 (93.85)	4 (6.15)
Male	57 (96.61)	2 (3.39)
Race			.526
White	75 (96.15)	3 (3.85)
Black	39 (92.86)	3 (7.14)
Other	4 (100)	0 (0)
Performance status			.880
0	15 (100)	0 (0)
1	73 (93.59)	5 (6.41)
2-3	30 (96.77)	1 (3.22)
Lesions			.209
1	102 (96.23)	4 (3.77)
≥2	16 (88.89)	2 (11.11)
Location			.835
LLL	19 (100)	0 (0)
LUL	29 (93.55)	2 (6.45)
RLL	28 (96.55)	1 (3.45)
RML	5 (100)	0 (0)
RUL	32 (91.43)	3 (8.57)
Media.	5 (100)	0 (0)
Lobes treated			.048*
1	112 (96.55)	4 (3.45)
2	6 (75)	2 (25)
Centrality			.213
Peripheral	57 (98.28)	1 (1.72)
Central	61 (92.42)	2 (7.58)
ICI receipt			.007*
No	68 (100)	0 (0)
Yes	50 (89.29)	6 (10.71)
Prior lung RT			.648
Yes	88 (95.65)	4 (4.35)
No	30 (93.75)	2 (6.25)
Prior RT courses			.859
0	65 (94.2)	4 (5.8)
1	35 (94.59)	2 (5.41)
≥2	18 (100)	0 (0)
Prior lung SBRT			1
Yes	44 (88)	6 (12)
No	6 (100)	0 (0)
Lung disease			1
No	106 (94.64)	6 (5.36)
Yes	12 (100)	0 (0)
Smoking history			1
No	22 (95.65)	1 (4.35)
Yes	95 (95)	5 (5)
Smoking current			.588
Yes	97 (94.17)	6 (5.83)
No	21 (100)	0 (0)
PTV sum (mean), cm3	59.39	138.76	.012*

Abbreviations: ICI = immune checkpoint inhibition; LLL = left lower lobe; LUL = left upper lobe; Media. = mediastinum; PTV = planning target volume; RLL = right lower lobe; RML = right middle lobe; RT = radiation therapy; RUL = right upper lobe; SBRT = stereotactic body radiation therapy.

^*Statistical significance (P < .05).

Discussion

Preclinical and early clinical data have described promising immune-mediated synergy between hypofractionated radiation and ICIs; several ongoing randomized studies are expressly testing that hypothesis.¹⁶ This study is the first to exclusively describe lung SBRT with concurrent ICI. Its median follow-up of 9.23 months compares favorably with similar series, which have ranged from 5.5 to 6.9 months.^13,17,18 As evaluated by technical parameters of SBRT plan quality, the desired conformality was achieved in the vast majority of patients. We found this approach safe overall, though the incidence of any- and high-grade pneumonitis appears elevated over reports from ICI alone and ICI + RT combinations. Interdigitated ICI administration and treatment of multiple lesions do not appear to confer excess risk of pneumonitis. However, caution should be exercised in the delivery of lung SBRT with ICI/ICI combination until prospective data mature, especially when done outside of the setting of a prospective trial or when in conjunction with known predictors of pneumonitis.

Because large randomized trials for ICI in the advanced NSCLC setting were not designed specifically to answer questions regarding RT safety, the risks of such combination therapy are not well understood. Patients undergoing any RT within 2 weeks of study start, or 30 Gy delivered within 6 months, have been excluded from advanced-stage immunotherapy studies.^19–22 Similarly, thoracic RT was not permitted in a phase 3 randomized study investigating the addition of atezolizumab to chemotherapy in extensive-stage small cell lung cancer.²³ Safety endpoints have been reported by a growing body of literature examining thoracic RT and ICI combinations.

However, the majority of experiences so far have excluded either SBRT or hypofractionated radiation or concurrent administration with ICI. The proportion of patients receiving SBRT, when applicable, has ranged from 23% to 24% as described by Shaverdian et al¹⁴ and von Reibnitz et al.¹⁸ Substantial heterogeneity exists among studies with regard to treatment approach and combination therapy pneumonitis risk. Emerging phase 1/2 data on the addition of concurrent ICI to definitive chemoradiation for stage III NSCLC has shown acceptable safety. Three of 30 patients experienced G2 or higher pneumonitis (n = 1 G3, n = 2 G2) with the addition of atezolizumab; G2+ pneumonitis was seen in 6 of 23 patients with the addition of pembrolizumab.^24,25 Retrospective comparative safety and toxicity data have been inconclusive. In a secondary analysis of Keynote-001, 98 patients with advanced NSCLC receiving pembrolizumab were analyzed to determine outcomes based on prior RT exposure. Receipt of thoracic RT was associated with increased risk of treatment-related pulmonary toxicity (13% vs 1%, P = .046). This was despite the majority (64%) of patients receiving palliative RT and the exclusion of those patients with a history of pneumonitis or autoimmune disease. Treatment-related G3 pneumonitis was low—4% after prior thoracic RT. As presented by Voong et al,²⁶ among 184 patients with advanced NSCLC being treated with anti-PD-1/PD-L1 therapy with prior radiation exposure, thoracic RT increased the risk of immune-related pneumonitis (25 vs 9%, P = .08).

Hwang et al¹² found no difference in immune-related AEs between patients with NSCLC who received prior thoracic RT versus those who received ICI alone. Additionally, there were differences with respect to any-grade (8.2% vs 5.5%) and G2 or higher pneumonitis (4.1% vs 3.3%). Notably, the median time between thoracic RT and ICI was 8.5 month (range, 0.1-69 months). A G2 or higher pneumonitis rate of 6.3% was reported when patients were treated with ICI either concurrently (within 30 days) or sequentially (within 180 days) with thoracic RT.¹⁸ Investigators from the University of Chicago prospectively evaluated the activity and safety of multisite SBRT with ICI.¹⁷ The majority of patients were treated with SBRT per NRG BR-001 dose and fractionation to 2 target lesions (94.5%) followed by pembrolizumab within 7 days after SBRT completion. Of 62 patients evaluable for toxicity, the risk of G3 or higher AEs by anatomic location was 8.3% for peripheral lung, 20% for central lung, and 10% for mediastinum. An institutional protocol of lung, liver, or adrenal SBRT (12.5 Gy × 4 or 6 Gy × 10) sandwiched by ipilimumab found only 2 instances of G3 pneumonitis, both occurring in a single patient (1.67%).¹³ Lastly, a small series of 6 patients with advanced NSCLC treated with lung SBRT followed by nivolumab noted a single instance of pneumonitis.²⁷

The incidence of immune-related pneumonitis as reported in meta-analyses has ranged from 0.4% to 4.3%, depending on primary histology, prior treatments, and ICI agent. The rates of any-grade and G3 pneumonitis were 3.6% and 1.1% for patients with advanced NSCLC.²⁸ Rates of any-grade and G3 pneumonitis in previously treated patients receiving PD-1/PD-L1 inhibition were 2.8% and 1%.²⁹ Conversely, large institutional data have questioned the rates seen in randomized studies. In a report on 205 patients from Johns Hopkins Hospital, including patients enrolled on clinical trials, the overall immune related-pneumonitis risk was 19%.³⁰ Fifty-three percent of cases were G3 or higher. Tamiya et al¹⁵ reported the development of radiation pneumonitis in 34 of 167 (20.3%) patients who received nivolumab; 50 patients had prior thoracic RT in this series.¹⁵ Rates reported here support these higher estimates, with 33.9% risk of any-grade pneumonitis and 10.7% incidence of G3 or higher. A multi-institutional analysis of 915 patients noted an elevated pneumonitis risk with ICI combinations over monotherapy (10 vs 3%, P < .01), supporting the trend observed here.³¹

Several aspects of the current study limit its wider generalizability. The study’s retrospective nature is an inherent limitation. Its sample size and incidence of high-grade pneumonitis precluded multivariable analyses. Moreover, despite multiple dosimetric endpoints being explored as potential predictors of pneumonitis, none were revealed to be statistically significant, presumably owing to the total population size and the relative rates of clinically meaningful pneumonitis. Patients differed in terms of receipt of prior lung radiation, number of prior RT courses, age, and pulmonary disease; regardless, none of these factors were identified as significant modifiers of pneumonitis risk in the SBRT-alone or SBRT + ICI cohorts. Though 14.3% of patients received treatment to more than 1 target lesion, those with meaningful overlap in dose as a result of multiple targets were underrepresented. There was substantial heterogeneity with regard to dose and fractionation, though 3 fractionation schemes represented the vast majority of cases (92.5%, 10 Gy × 5, 12 Gy × 4, and 18 Gy × 3).

The optimal integration of highly conformal ablative RT and ICI has yet to be determined; however, a growing body of translational and human data support their synergy. An elegant study by Formenti et al³² showed that RT induced a rapid expansion of T-cell clones specific for an RT-upregulated neoantigen in a patient with NSCLC receiving ipilimumab.³² Data regarding the optimal sequencing and dosing of these 2 treatment modalities have been recently reviewed.³² Preclinical data supported an optimal fraction size of less than 12 Gy due to TREX-1 induction at higher doses and attenuation of the immune response.³⁴ Additionally, very close sequencing of RT and ICI is likely optimal because intratumoral CD8 T-cell frequencies peak at 5 days post-RT, at which point the highest effector to Treg ratio can be seen.³⁵

The number of ongoing prospective studies highlights the rapidly growing research interest. PEMBRO-RT is a phase 2 randomized enhancing systemic tumor response with the addition of 8 Gy × 3 SBRT to a single target lesion before pembrolizumab in advanced NSCLC (NCT02492568). A phase 1 study testing combination sequencing is assigning patients with 5-fraction SBRT to concurrent, induction, or sequential delivery of atezolizumab (NCT02400814). Other studies are interested in novel combinations, including SBRT and gene therapy followed by nivolumab (NCT02831933), FLT3 ligand immunotherapy with SBRT (NCT02839265), or CD137 or CSF-1R inhibitors with nivolumab and SBRT (NCT03431948).

Pneumonitis is an overlapping toxicity in patients receiving immune modulatory agents and thoracic radiation. The prevalence of clinically significant pneumonitis was approximately 10%, as reported here. This risk, among known risk factors, should be considered when assessing immunotherapy patients for SBRT candidacy. Though no statistically significant dosimetric parameters were identified here, efforts nonetheless should be made to minimize lung exposure. Additional studies in a larger patient cohort with homogeneity in treatment approach and patient characteristics may better describe the interplay of clinical and dose-volume parameters, aiding the development of clinical practice and treatment planning guidelines.

Conclusions

The risk of pneumonitis may have been underappreciated for this patient population based on prior reports. Increased recognition of and close observation for early symptoms is advocated for all who participate in the management of patients receiving ICI, especially when administered in close proximity to the timing of ablative RT.