EVALUATION OF RADIATION-INDUCED PLEURAL EFFUSIONS AFTER RADIOTHERAPY TO SUPPORT DEVELOPMENT OF ANIMAL MODELS OF RADIATION PNEUMONITIS

Abstract

Introduction:

Not all animal models develop radiation-induced pleural effusions (RIPEs) as a form of radiation-induced lung injury (RILI). Such effusions are also not well-characterized in humans.

Objectives:

The purpose of this study is to identify occurrences of RIPE in humans, provide justification for development of relevant animal models, and further characterize its risk factors in cancer patients. We also aim to identify dose thresholds for cardiopulmonary toxicity in humans to shed light on possible pathogenic mechanisms for RIPEs.

Methods:

We carried out a retrospective review of medical records of 96 cancer patients receiving thoracic irradiation (TRT) at our institution.

Results:

53 (55%) patients developed a new pleural effusion post TRT; 18 (19%) had RIPE. 67% developed RIPE ipsilateral to the site irradiated, none developed ‘contralateral only’ effusions. Median time to development was 6 months (IQR; 4–8 months). Of 18, 8 patients (44%) had concomitant asymptomatic (radiographic only) or symptomatic radiation pneumonitis and pericardial effusion. Dosimetric factors including combined and ipsilateral mean lung dose (MLD) were significantly associated with increased risk of RIPE. Angiotensin converting enzyme inhibition, steroids or concurrent chemotherapy did not modify incidence of RIPE.

Conclusions:

Our results substantiate the occurrence, and incidence, of RIPEs in humans. In cancer patients, a median time to development of effusions around 6 months also supports the onset of RIPEs concurrent with radiation pneumonitis. Future work needs to include large populations of cancer survivors in whom delayed RIPEs can be tracked and correlated with cardiovascular changes in the context of injury to multiple organs.

INTRODUCTION

Animal model development is a major focus of the radiation countermeasure program at the National Institute of Allergy and Infectious Diseases (NIAID) and the Biomedical Research and Development Authority (BARDA). Countermeasures need to be tested in animal models that reflect radiation injuries in humans, in order to be approved by the Food and Drug Administration (FDA). Animal model development must evaluate key organ-specific sequelae of radiation toxicity and define radiation dose with incidence for acute and delayed injuries (MacVittie 2020).

The lungs are sensitive to ionizing radiation and may develop radiation-induced lung injury (RILI), a term used synonymously with acute radiation pneumonitis (RP) and/or chronic pulmonary fibrosis (Van Dyk et al. 1981, Deas et al. 2017, Simone et al. 2017). Animal models reveal that whole thorax lung irradiation (WTLI) has far reaching effects beyond just target cell damage, including vascular injury, pulmonary hypertension, and alterations in the physiological interaction between heart and lungs (Ghosh et al. 2009, Ghobadi et al. 2012b, Medhora et al. 2015, Jacobs et al. 2019). Preclinical models show that radiation exposure is associated with development of radiation induced pleural effusions (RIPEs), i.e., in rats (Medhora et al. 2015), some strains of mice (Jackson et al. 2010, 2011) and nonhuman primates (NHPs) (Garofalo et al. 2014). Data from these models also help advance our understanding of how genetic susceptibility (Jackson et al. 2010), risk and mitigating factors, including radiation dose thresholds, contribute to RILI (Medhora et al. 2012, 2019), including RIPEs (Jackson et al. 2010, Garofalo et al. 2014), in animals.

There is controversy however, regarding the development of RIPEs in humans. This issue must be resolved in order to select appropriate animal models for countermeasure development. In 2010, researchers cautioned against the use of C57BL/6 mice for radiation countermeasure studies as they developed RIPEs; a consequence of radiotherapy then considered rare in humans (Jackson et al. 2010). However, RIPEs have been described in humans, as early as in the 1950s in patients receiving breast radiotherapy (Bachman 1959) and later as case observations (Whitcomb 1971, Rodríguez-García et al. 1991, Morrone et al. 1993). What is clear is that compared with animal models, there are limited data describing the incidence, clinical characteristics, risk factors (including dose thresholds) that are associated with development of RIPEs in humans.

Since it is neither ethical nor feasible to conduct relevant irradiation toxicity studies in humans, the human response to thoracic irradiation (TRT) must carefully be investigated using data available from clinic. Some work has helped in this direction. For instance, it’s been shown that 18% of patients receiving TRT for esophageal cancer developed pleural effusions exceeding grade 2 toxicity and mediastinal radiation field width > 8 cm poses a significant risk for RIPEs (Fukada et al. 2012). Similarly, Zhao et al. have also recently demonstrated that RIPEs occur frequently in humans (29%) and dosimetric factors such as V5 (volume of lung receiving ≥ 5 Gy) are important risk factors for developing RIPEs (Zhao et al. 2017). With just a handful of human studies on this important topic, further confirmation and a more comprehensive characterization is still needed to understand pathogenic pathways responsible for RIPEs.

Furthermore, because humans cannot be administered toxic doses of radiation to study RILI, animals that demonstrate similar pathogenic pathways and closely emulate RILI in humans must be used to study effects. Similarly, the Food and Drug Administration (FDA) must approve and stockpile countermeasures for radiation toxicity based on results derived from irradiated animals rather than human clinical trials (Animal Rule). For these reasons, detailed observations from humans are important in ‘fine-tuning’ the selection of preclinical models that can then be developed to conduct large scale studies to study radiation effects.

The purpose of this observational study is three-fold. First, by carefully reviewing clinical data on cancer patients receiving TRT, and excluding those with other possible etiologies, we identify occurrence, and substantiate incidence, of RIPE in humans with a high level of confidence. Second, by elucidating the clinical milieu in which these effusions arise (i.e., concomitant serositis, cardiovascular deterioration requiring diuresis) we aim to shed light on potential pathogenic mechanisms that contribute to the development of RIPEs in humans. Third, by identifying patients who develop RIPEs, we delineate treatment effects, including laterality and dose thresholds, that predict risk for cardiopulmonary toxicities. The role of radiation injury to the lung in the presence of concomitant multiple organ injury as described in animals (MacVittie 2020) is not investigated here. However, the current study in patients does relate to localized RILI, as also described in mice (Jackson et al. 2010, 2011), rats (Medhora et al. 2015) and nonhuman primates (Garofalo et al. 2014).

MATERIALS AND METHODS

Our retrospective studies were approved by the Institutional Review Board at our institution.

Patient eligibility

Case records of lung or esophageal cancer patients who received TRT between December 2011 to July 2015 at a single medical center were reviewed. Patients fulfilling the following criteria were included: a) pathologically confirmed cancer (non-small cell, small-cell lung or esophageal), b) received TRT with or without chemotherapy and c) had imaging available (chest x ray (CXR), computed tomography (CT) scan, positron-emission tomography-computed tomography (PET/CT). Clinical staging was classified according to the American Joint Commission on Cancer staging system (7^th edition) (Goldstraw et al. 2007).

Patient demographic, disease and treatment details and follow-up

Electronic health records were reviewed by a pulmonologist. An additional review was performed by a second physician or as a multidisciplinary review in cases of uncertainty. Baseline socio demographic information, patient-factors (age, gender, race, smoking status, long-term oxygen therapy (LTOT), angiotensin converting enzyme inhibitor (ACEi) use), cancer details (histological subtype, clinical stage, chemotherapy, and radiation treatments), history of prior surgery, chest irradiation were recorded. When available, baseline & follow-up cardiopulmonary evaluations (pulmonary function testing, transthoracic/transesophageal echocardiogram, or coronary catheterization) were tracked. All available imaging was reviewed for underlying parenchymal disease, emphysema, RP, fibrosis, new pleural or pericardial effusions. Chemotherapy and radiation treatment plans were recorded. All patients were followed until the last visit or death.

Patients received conventionally fractionated TRT (1.8 or 2.0 Gy/fraction) using 6 MV (vast majority) or 15, 18 MV photon beams (small minority). Most patients received ≥ 60 Gy in 30 fractions over 6 weeks. All underwent three-dimensional conformal radiotherapy or intensity modulated radiotherapy (3D CRT or IMRT) with CT simulation (slice thickness of 2.5–4.0 mm). The dose volume histogram (DVH) parameters analyzed included mean lung dose (MLD) and volume of total lung receiving 5, 20, and 30 Gy (V5, V20, and V30, respectively). DVH parameters were calculated individually for both right and left lung, total combined lung volume, and total lung volume excluding the treatment prescription target volume (PTV). Records from follow up with primary oncologist/radiation-oncologist were reviewed for treatment updates, change in symptoms, physical or functional status (i.e., ECOG status, new oxygen requirement). Adverse events were graded retrospectively according to the National Cancer Institute-Common Toxicity Criteria (CTCAE) version 5.0 (US Department of Health and Human Services, National Cancer Institute 2017).

Evaluation of pleural effusion

All imaging was reviewed to identify patients with new pleural effusions (including loculated effusions) post TRT. Once a patient with a new pleural effusion was identified, we performed a detailed chart search of plausible clinical explanations for the occurrence of a new pleural effusion (i.e., positive cytology in pleural fluid signifying progressive malignancy, radiographic progression of malignancy, concomitant pneumonia and parapneumonic effusion or decompensated heart failure). After exclusion of possible alternate clinical explanations, patients were defined as ‘pleural effusions of an undetermined etiology’. In the context of prior TRT these were further defined as ‘radiation induced pleural effusions or RIPE’.

Definitions

Radiation induced pleural effusions (RIPEs)

We selected the validated model to classify effusions radiographically as ‘small’ (< 25% of hemithorax), ‘moderate’ (25–50% of hemithorax) or ‘large’ (50–75% or 75–100%) based on anteroposterior distance (AP distance) on cross-sectional CT view (Moy et al. 2013). Severity was graded according to the CTCAE v5.0 criteria (Grade 1: ‘asymptomatic’/clinical or diagnostic observation only/no intervention), Grade 2: ‘symptomatic’/diuretics or limited therapeutic thoracentesis, Grade 3: respiratory distress or hypoxia requiring chest tube or pleurodesis, Grade 4: life threatening respiratory or hemodynamic compromise requiring intubation/urgent intervention, Grade 5: death) (US Department of Health and Human Services, 2017). ‘Time to RIPE’ was estimated from last day of TRT to the date at which pleural effusion was first identified on follow up imaging.

Radiation pneumonitis (RP)

‘Radiation pneumonitis’ was identified by the identification of imaging findings (ground-glass opacities conforming within radiation ports) with or without compatible symptoms (nonproductive cough, dyspnea on exertion, low grade fevers or hypoxemia) developing within 12 weeks of TRT. Severity was graded using the CTCAE v5.0 criteria (Grade 1: ‘asymptomatic, requiring no treatment, Grade 2: ‘symptomatic’ requiring steroids or diuretics, Grade 3: ‘symptomatic’ requiring supplemental oxygen, Grade 4: life-threatening respiratory compromise requiring mechanical ventilation, Grade 5: death) (US Department of Health and Human Services, 2017).

Statistical analysis

Data analysis was performed by using Statistical Package for Social Sciences software, version 19.0 (SPSS Statistics, IBM). Categorical variables were summarized using frequencies and percentages and continuous data are presented as medians (interquartile ranges (IQR)). The Mann Whitney U (Wilcoxon rank sum) test was used for continuous variables and the χ2 test or the Fisher exact test for categorical variables. Variables with a p value < 0.05 in the univariate analysis were entered into multivariate logistic regression analysis to identify independent risk factors with RIPE. All p values < 0.05 are considered statistically significant.

RESULTS

Baseline demographics

Records of 100 patients were reviewed (Table 1). Four were excluded; one without imaging, one with concomitant lung and esophageal malignancy (difficult to ascertain dose etc.) and two with recurrent cancer (prior irradiation).

Table 1.

Baseline characteristics of all patients

Baseline characteristics	Total no. of patients (N = 96) (%)
Gender
Male	60 (63.5)
Female	36 (37.5)
Age at diagnosis of cancer (median, IQR)	65.5 years (58–72 years)
No. of patients alive today *	60 (62.5)
Biometrics (median, IQR)
Weight (kg)	77 (63.3–91)
Height (meters)	1.7 (1.6 – 1.8)
BMI (kg/m2)	27 (23 – 32)
Ethnicity
White (Caucasian)	73 (76)
Black (African American)	17 (18)
Native American or Indian	3 (3)
Other (Asians & Hispanics)	3 (3)
Primary site of cancer
Lung	69 (72)
Esophageal	26 (27)
Colon metastatic to lung	1 (1)
Stage of Cancer
Stage I	12 (13)
Stage II	11 (12)
Stage III	56 (58)
Stage IV (metastatic)	14 (15)
Recurrence of cancer	5 (5)
Ever smokers	84 (88)
Current active smoker (at time of TRT)	37 (39)
ACE inhibitors
Use of ACE inhibitor (before or during TRT)	24 (25)
Duration of ACE inhibitor (before/during TRT) (mean ± SD)	7.5 ± 18.9 months
Use of oxygen (at time of TRT)	14 (14.6)
Steroid use prior to TRT	19 (19)
Baseline cardiac dysfunction	39 (41)
Congestive heart failure (systolic, diastolic or both)	22 (23)
Valvular heart disease	11 (11)
Arrhythmias	10 (10)
Coronary artery disease	36 (38)
Baseline pulmonary dysfunction	78 (79)
Restrictive impairment (ILD, asbestosis etc.)	10 (10)
COPD/Asthma (Obstructive impairment)	64 (67)
Pulmonary hypertension	3 (3)
Pleural effusions (present before TRT)	5 (5)
Pulmonary embolism	4 (4)
Obesity/Obstructive sleep apnea	7 (7)
Baseline pulmonary function testing
Absolute FEV1 (L) (median, range)	2.0 (0.6 – 3.7)
Absolute FVC (L) (median, range)	3.1 (1.5 – 4.9)
FEV1/FVC (median, range)	65 (31 – 81)
Absolute TLC (L) (median, range)	5.5 (2.3 – 8.7)
Absolute DLCO (mL/min/mmHg) (median, range)	13.6 (3.1 – 25.7)
Concurrent chemotherapy	85 (89)
Combined MLD (cGy) (median, IQR) †	1349 (998 – 1710)
Total duration of TRT (median, IQR)	41 days, (32.5–44)

^†Complete TRT data was available on 89 patients

^*Till completion of data collection in July 2017.

CI = Confidence Intervals, cGy = centigray. IQR = Interquartile range, MLD = mean lung dose

Of 96 patients, 60 (63%) were male with a median age of 65.5 years (IQR; 58–72) at diagnosis and 60 were alive by end of data collection (July 2017). Median follow-up was for 3.3 years (IQR; 2.6 – 4.3) after diagnosis. Our study population included Whites (76%), African Americans (18%) and others (6%). Lung was the commonest cancer site (71%) followed by esophagus (27%). 58% had stage III disease, 29.5% had prior surgery and 25% were currently receiving, or had recently used, angiotensin converting enzyme inhibitors (ACEi). 41% had baseline cardiac dysfunction (commonest; congestive heart failure) and 79.6% had a preexisting pulmonary condition (commonest; obstructive ventilatory impairment (COPD/Asthma)). Baseline lung function was mildly reduced; median FEV₁/FVC of 65% (range: 31–81%), median FEV₁ of 2.0 L (range: 0.6 – 3.7 L).

Thoracic irradiation (treatment plans and general effects)

TRT data were available for 89 patients. Median MLD delivered to combined lungs was 1349 cGy (IQR; 998 – 1710), right lung was 1193 cGy (IQR; 852–2113) and left lung was 995 cGy (IQR; 644–1643). Median TRT duration was 41 days (IQR; 32.5 – 44). 62 patients (64.6%) developed either asymptomatic (radiographic only) or symptomatic RP. 19 patients (32.8%) developed ≥ Grade 2 pneumonitis requiring intervention a median of 2 months (range; 1–12) after TRT.

Pleural effusions

Of 96 patients, 53 (55%) developed a new effusion (Table 2). Of these, 3 had parapneumonic effusions, 4 had decompensated cardiac failure, 8 had progressive malignancy (1 with positive pleural fluid cytology, 7 with radiographic progression with a new effusion suggestive of metastasis), 1 had postoperative chylothorax, 5 had ‘self-resolving’ perioperative effusions and 4 had effusions before TRT. In addition, 10 had multiple factors contributing to the development of an effusion (i.e., combination of pulmonary embolism, decompensated cardiac function and/or superimposed infection) (Table 2). Overall, 35 of 53 patients had an alternate plausible explanation for a new effusion. Others were determined to have ‘effusions of an underdetermined etiology’ or suspected radiation induced pleural effusions (RIPE).

Table 2.

Clinical scenario for the development of pleural effusions post-irradiation

Baseline characteristics	Total patients (N = 96)
Pleural Effusion detected	53 patients
Pleural effusion secondary to worsened cardiac function (Evidence of concomitant worsening cardiac function, diuresis prescribed with improvement etc.)	4 patients
Pleural effusion secondary to pulmonary infection	3 patients
Pleural effusion secondary to progressive malignancy
Pleural fluid cytology positive	1 patient
Evidence of concomitant disease progression elsewhere	7 patients
Chylothorax (post-operative)	1 patient
Peri- or post-surgery (peri-esophagectomy/post pneumonectomy etc.)	5 patients
Effusions were present before TRT	4 patients
Multifactorial etiologies
Pulmonary embolism/worsened cardiac function	2 patients
Concomitant infection and worsened cardiac function etc.)	8 patients
*Pleural effusions of an undetermined etiology
Radiation-induced pleural effusions (RIPE)	18 patients

Radiation induced pleural effusions

18 (19%) patients developed RIPE, 13 (72%) were symptomatic. 12 (67%) developed effusions ipsilateral to the site irradiated while three had bilateral effusions. All with bilateral effusions had received TRT in proximity to the mediastinum due to a central location of their primary tumor. None developed ‘contralateral only’ effusions to the site irradiated.

Age, gender, body mass index, race/ethnicity, oxygen or ACEi use, smoking status, tumor origin, stage, histology, survival since diagnosis, preexisting cardiopulmonary disease or baseline cardiopulmonary function were not significantly associated with RIPE. End of treatment lung function as measured by FEV₁/FVC was higher in those with RIPE (median 68 vs. 57, p = 0.02). 88.5% received concurrent chemotherapy and this did not influence RIPE development (p = 0.38).

14/18 developed ‘small’ effusions (64% symptomatic) whereas 4/18 developed ‘moderate’ effusions (all symptomatic). Three required tube-thoracostomy. Median time to RIPE was 6 months. 9 of 18 patients with RIPE developed a concomitant pericardial effusion and this rate of occurrence was higher than in those without RIPE ((19 of 78), 50% vs 24.4%, p = 0.031)). Similarly, 16 of 18 patients with RIPE also developed RP (any grade) and this occurrence was also higher than in those without RIPE (88.9% vs. 59%, p = 0.017). Incidence of ≥ Grade 2 pneumonitis was not statistically significant in those with or without RIPE (27.8% vs. 18.9%, p = 0.405). Of note, 8 of 18 (44.4%) patients had concomitant (symptomatic or ‘asymptomatic’ (radiographic only)) evidence of all three; RIPE, pneumonitis and new pericardial effusion (Tables 3 and 4).

Table 3.

Characteristics of patients with radiation-induced pleural effusion (RIPE)

Patient w/ RIPE	Baseline lung disease	Time since TRT	Chemotherapy Received	Side of RT	Side of effusion	Size of pleural effusion	Clinical severity*	New pericardial effusion	Radiation pneumonitis
Patient 1	COPD	6 months	Carboplatin/taxol	Right	R > L	< 25%	Asymptomatic	No	Yes
Patient 2	Combined emphysema/fibrosis	1 month	Carboplatin/taxol	Left ↔	Left	< 25%	Asymptomatic	Yes	Yes
Patient 3	OSA	17 months	Carboplatin/taxol	Central/midline	Bilateral	25–50%	Symptomatic	No	No
Patient 4	COPD	6 months	Carboplatin/Abraxane	Right ↔	Right	< 25%	Asymptomatic	No	Yes
Patient 5	OSA	2 months	Carboplatin/taxol	Central/midline	Bilateral	< 25%	Asymptomatic	Yes	Yes
Patient 6	COPD	2 months	Carboplatin/Abraxane	Right ↔	Right	25–50%	Symptomatic	No	Yes
Patient 7	COPD/Asthma/OSA	9 months	Carboplatin/taxol	Left ↔	Left	< 25%	Symptomatic	No	Yes
Patient 8	COPD	5 months	Carboplatin/taxol	Right ↔	Right	< 25%	Asymptomatic	No	Yes
Patient 9	COPD/OSA	6 months	Carboplatin/taxol	Right ↔	Right	< 25%	Symptomatic	Yes	Yes
Patient 10	0	7 months	DFOX	Central/midline	Left	< 25%	Symptomatic	No	Yes
Patient 11	History of bilateral pulmonary embolism	8 months	Carboplatin/taxol	Central/midline	Left	< 25%	Symptomatic	Yes	Yes
Patient 12	0	1 month	Carboplatin/taxol	Right ↔	Right	< 25%	Symptomatic	Yes	Yes
Patient 13	COPD	12 months	None	Right ↔	Right	< 25%	Symptomatic	Yes	Yes
Patient 14	0	18 months	Carboplatin/taxol	Central/midline	Bilateral	< 25%	Symptomatic	No	Yes
Patient 15	COPD	7 months	Carboplatin/taxol	Left ↔	Left	< 25%	Symptoms w/ respiratory distress/hypoxia	No	Yes
Patient 16	COPD / Pulmonary Embolism	8 months	Carboplatin/taxol	Left ↔	Left	< 25%	Symptomatic	Yes	Yes
Patient 17	COPD	4 months	Carboplatin/taxol	Right ↔	Right	25–50%	Symptoms w/ respiratory distress/hypoxia	Yes	No
Patient 18	COPD	4 months	Carboplatin/taxol	Left ↔	Left	25–50%	Symptoms w/ respiratory distress/hypoxia	Yes	Yes

^*For pleural effusion, grade of clinical severity as follows; Asymptomatic; clinical or imaging evidence only, no intervention needed, Symptomatic, intervention needed (diuresis or limited thoracentesis), Symptoms with respiratory distress and hypoxia (surgical intervention such as a chest tube needed)

Table 4.

Comparison of patients who develop RIPE vs. all others

Patient Specific Factors	‘All Others’ (n = 78)	RIPE (n = 18)	p-value
Age (at diagnosis) (median, range)	66 (45–88)	63.5 (47–88)	0.87 W
BMI (kg/m2) (median, range)	26 (17–41)	28 (15–61)	0.35 W
Gender
Females	29 (37.2)	7 (38.9)	0.89 C
Males	49 (62.8)	11 (61.1)
Ethnicity			0.19 C+
Caucasian	61 (78.2%)	12 (66.7%)
African American	14 (17.9%)	3 (16.7%)
Filipino	1 (1.3%)	1 (5.6%)
Native American/Indian	2 (2.6%)	1 (5.6%)
Asian & Hispanic	0 (0%)	1 (5.6%)
Baseline FEV1 (median, range)	2.0 (0.9 – 3.7)	1.8 (0.6 – 3.5)	0.74 W
Baseline FEV1/FVC (median, range)	65 (31– 81)	68 (35 – 81)	0.93 W
End of Rx FEV1 (median, range)	1.5 (0.7 – 3.0)	2.2 (1.1 – 3.7)	0.09 W+
End of Rx FEV1/FVC (median, range)	57 (23– 76)	68 (59 – 77)	* 0.02 W
Pre-existing cardiac disease	31 (39.7%)	8 (47.1%)	0.58 C
Use of an ACEi before TRT	17 (21.8%)	7 (38.9%)	0.23 C+
ACEi duration (mean months) ± SD)	6.5 ± 17.5	12 ± 24.4	0.20 W
Use of supplemental oxygen	11 (14.1%)	3 (16.7%)	1.00 C+
Use of steroids before TRT	16 (20.5%)	3 (16.7%)	0.76 C+
Smoking status
Active smoker	32 (41%)	5 (27.8%)	0.29 C
Pack years (median, range)	40 (0 – 154)	39 (0 –120)	0.64 W
Cancer-Specific Factors
Stage of Cancer			0.14 C
I	11 (14.1%)	1 (5.6%)
II	6 (7.7%)	5 (27.8%)
III	47 (60.3%)	9 (50.0%)
IV	12 (15.4%)	2 (11.1%)
Cancer Origin
Primary Lung	56 (71.8%)	13 (72.2%)	0.97 C
Primary Gastrointestinal Tract	22 (28.2%)	5 (27.8%)
Treatment Specific Factors (TRT & Chemotherapy Regimens Used)	‘All Others’(n = 78)	RIPE (n = 18)	p-value
Concomitant chemotherapy	68 (87.2%)	17 (94.4%)	0.38 C
Other Clinical Observations
Concomitant pericardial effusion	19 (24.4%)	9 (50.0%)	*0.031 C
Concomitant radiation pneumonitis †	46 (59%)	16 (89%)	*0.017 C
Clinical severity of XRT pneumonitis			0.41 C
Didn’t develop OR asymptomatic (no intervention)	60 (76.9%)	13 (72.2%)
Grade 2 or higher (requiring intervention)	14 (17.9%)	5 (27.7%)

^*Statistically significant

⁺Exact test

^TT-test

^WWilcoxon rank-sum test

^Cchi-square test

^†Asymptomatic (radiological only) and/or symptomatic (symptoms and radiological) radiation pneumonitis

Lung radiation doses were higher in those who developed RIPE. Logistic regression analyses revealed a weak but positive association for predicting RIPE based on combined MLD (odds ratio for RIPE; 1.136 (95% CI 0.998, 1.293) per 100 cGy, p = 0.046 (likelihood ratio test)). Multivariable logistic regression analysis to predict RIPE based on MLD delivered to single, ipsilateral lung revealed odds for developing RIPE were 1.069 (95% CI 1.002, 1.139) per 100cGy for the right lung (p = 0.033) and 1.074 (95% CI 1.008, 1.143) per 100cGy for the left lung (p = 0.026). These models indicate a positive association between combined and ipsilateral MLDs and RIPE (Table 5 and Fig. 1). No other factors were statistically significant after controlling for MLD.

Table 5.

Comparison of radiation dose between patients with RIPE versus ‘All Others’

Combined Lung Dose	‘All Others’	RIPE	p-value
Median (range) (cGy)	1294 (414 – 2190)	1528 (745 – 2626)	0.064 W
Mean ± SD (cGy)	1307 ± 424	1540 ± 463	* 0.048
V5 (%) ^	62.37 ± 16.22	67.08 ± 15.41	0.28 W
V20 (%) ^	33.70 ± 10.37	36.19 ± 8.83	0.36 W
V30 (%) ^	24.98 ± 7.96	28.11 ± 6.96	0.14 W
Logistic Regression Analysis
	Mean Lung Dose	OR+ per 100 cGy	95% CI	p-value
Combined	1352 cGy	1.136	(0.99 – 1.29)	* 0.046
Right Lung	1513 cGy	1.069	(1.002 – 1.139)	* 0.033
Left Lung	1298 cGy	1.074	(1.008 – 1.143)	* 0.026

^(W)Wilcoxon rank-sum test

(OR⁺); Odds Ratio for RIPE development

^(*)statistically significant

^{(^)}% combined lung volume irradiated with 5 Gy, 20 Gy or 30 Gy, respectively

Figure 1:

Comparison of percentage volume of lung for those who developed RIPE vs. all others. The % of lung that received ≥ 5 Gy, ≥ 20 Gy or ≥ 30 Gy are all likely to be higher for patients who develop RIPE.

DISCUSSION

In this retrospective study we present a review of the clinical characteristics of RIPEs in humans. First, by careful exclusion of other etiologies, we found 19% patients developed RIPE (72% were symptomatic). 67% developed ipsilateral effusions while 3, who received TRT close to the mediastinum, developed effusions bilaterally. None developed ‘contralateral only’ effusions. Median time to RIPE was 6 months (range 1–18). Our findings agree with those of Zhao et al. who found similar incidence (29%; 22% new and 7% increase in preexisting effusions) and median time to development for RIPE (3.7 months; range 0.6–18) (Zhao et al. 2017). In their study, majority patients were symptomatic (80%) – much as in ours.

Second, we describe further that RIPEs in humans tend not to occur contralaterally and are always ‘small or moderate’ in size, never exceeding > 50% hemithorax. Our results also show that patients who develop RIPE are significantly more likely to exhibit (asymptomatic or symptomatic) pneumonitis (89%) and pericardial effusions (50%), with all three anomalies observed in just under half cases (44%). Indeed, literature addressing early cardiac toxicity after TRT also suggests that ‘trace’ or ‘small or moderate’ sized pericardial effusions develop in patients at a median interval of 6–12 months (Borkenhagen et al. 2016).

In the context of available information from animals and cancer survivors (Whitcomb 1971, Rodriguez-Garcia et al. 1991, Morrone et al. 1993, ATS Guideline 2000, Medhora et al. 2015, Zhao et al. 2017) our observations lead us to consider several important aspects in the pathogenesis of RIPE. Given laterality, size and evidence of surrounding inflammation in our study, we speculate that RIPEs in humans represents more a locoregional ‘pleuritis’ or serositis rather than global effects of cardiovascular remodeling (influencing hydrostatic and pleural fluid pressures). Timelines here appear to fit those of early complications of TRT, i.e. pleuritis (typically occurring 6 weeks to 6 months after TRT) (ATS Guideline 2000).

Third, based on logistic regression analyses, we find that lung radiation doses were higher in those who developed RIPE vs. others. Specifically, combined MLD, a well-established dosimetric risk factor for pneumonitis (Graham et al. 1999, Zhang X-J et al. 2012) and indeed, ipsilateral MLDs are significant factors associated with RIPE in our study. We also further define combined and ipsilateral lung dose thresholds for developing RIPE (Table 5). Zhao et al. have previously demonstrated that combined lung V5 > 41% is a significant factor predicting RIPE in patients, particularly ‘symptomatic RIPE’ (Zhao et al. 2017). In our study, although combined lung V5, V20, V30 (percent volume of lung receiving ≥ 5 Gy, 20 Gy and 30 Gy, respectively) were higher for those who developed RIPE vs. those who did not, these data did not achieve statistical significance. A vast majority of irradiations were performed using 6 MV photon beams (average photon energy of ~2 MeV (1/3 the nominal energy)). Our observations along with those of Zhao et al. (Zhao et al. 2017) emphasize that below certain thresholds, both ‘radiation dose’ (combined and ipsilateral MLD) and ‘percentage volume of lung irradiated’ (V5) maybe important risk factors in the development of RIPE. This holds important implications for future models of dose planning for radiotherapy.

While our results support prior observations in humans (Fukada et al. 2012, Zhao et al. 2017), they also highlight important comparisons in radiation effects between animals and humans. For instance, we observe that a far greater proportion of animals develop effusions much sooner after TRT, than humans. As many as 76% Wistar rats develop RIPE by 6 weeks after 15 Gy WTI (Medhora et al. 2015) and 75% macaques develop effusions by day 30 after 12 Gy WTI (Garofalo et al. 2014). These differing observations may be explained by the strict limitation of dose and lung volumes used in patients as compared with the total volume of lungs subjected to > 9 Gy in animals (Jackson et al. 2010, 2011, Medhora et al. 2015, Garofalo et al. 2014). This variation in dose delivery, and effect, is also relevant to other forms of RILI; a single dose between 10 – 14 Gy of ionizing radiation administered to whole lung is sufficient to cause lethal pneumonitis in most species including humans (Van Dyk et al. 1981, Garofalo et al. 2014, Deas et al. 2017) whereas, as in our study, a localized dose of ≥ 13.5 Gy (MLD) delivered over several weeks appears to increase risk for RIPE in humans. While both animal models and human data add invaluable information, it is conceivable that differences in radiation delivery methods (i.e., single large WTI vs. ‘conformational’ or mapped ‘radiotherapy’ (3D CRT) delivered in fractions) are major factors defining the clinical picture. Additionally, although there is evidence from animal models that RIPEs may be mediated by a complex, dose and time-dependent multiorgan injury, as opposed to just a localized effect, information on these complex effects, is limited from human data. Indeed, histological observations from rat models report exuberant pleural fluid inflammation (macrophages, mast cells, ‘exudates’) (Jackson et al. 2010), Medhora et al. 2015) befitting intense ‘pleural inflammation’, much as the ‘early’ effects seen in humans in our study. But animal models also show quite convincingly that cardiopulmonary vascular remodeling does indeed occur in response to sublethal WTI; as heralded by a sustained reduction in arterial distensibility and vessel density (‘drop out’), persistently elevated pulmonary vascular resistance (PVR), right ventricular strain, impaired left ventricular diastolic filling, and diminished cardiac output (reduced to 30% of controls) (Ghosh et al. 2009, Ghobadi et al. 2012a, 2012, Medhora et al. 2015, Jacobs et al. 2019) – delayed changes that also predispose to developing delayed RIPEs. These effects could be related to the fact that in animal models, the whole heart and lungs are also within field of radiation. Nevertheless, they raise important questions of potential cardiovascular remodeling post TRT in humans. Although we tracked data on patients for a median of 3.3 years after diagnosis, it is possible that these are effects are even more delayed and/or apparent only if specific serial testing is performed (echo, right heart catheterization or N-terminal pro-B-type natriuretic peptide). The volume of the heart irradiated, and the dose of radiation administered needs also to be considered closely to detect potential delayed cardiovascular effects, highlighting an important area for large, population based future work.

Finally, while our study supports and substantiates the finding that pleural effusions occur frequently in humans as a consequence of radiotherapy (Zhao et al. 2017) and challenges the paradigm that RIPEs occur rarely in irradiated patients (Jackson et al. 2010, 2011), there remains the question of selecting animal models that adequately recapitulate the pathogenesis caused by a threat agent in humans (in this case ‘irradiation’). Researchers have actively pursued murine models of irradiation for the discovery of mitigators of lung injury (Jackson et al. 2010, 2011, Rabender et al. 2016) and RIPE has been described in nonhuman primates (Garofalo et al. 2014), rats (Medhora et al. 2015) and mice (Jackson et al. 2010, 2011). Since relevant irradiation toxicity studies cannot be performed in humans, our observations that irradiated humans develop RIPE and RIPE is accompanied by pneumonitis, are important and timely. These results should inform the choice of relevant preclinical models for ongoing research to develop countermeasures for a radiological attack or nuclear accidents, under the FDA Animal Rule.

Our data are limited in several respects. This study includes patients given radiation to only a partial volume of the lung without exposure of multiple organs, the latter scenario most likely to occur in a radiological attack (MacVittie 2020). In this retrospective review we did not specifically test to detect new onset congestive heart failure or pulmonary hypertension after TRT and cannot address cardiopulmonary vascular remodeling in humans. We also did not find any association with underlying patient or disease factors that have previously been suggested to impact RIPE, i.e., gender, ethnicity, baseline lung function or cancer stage (Zhao et al. 2017) nor specific treatment related factors such as concurrent chemotherapy, use of ACEi or steroids (Garofalo et al. 2014, van der Veen et al. 2015, Zhao et al. 2017). This lack of association may be because our sample size was not large enough to detect these differences.

CONCLUSION

RIPEs occur in cancer patients and tend to develop as ipsilateral, small to moderate sized effusions and MLD (combined and ipsilateral) are important determinants for development. RIPE is accompanied by (asymptomatic or symptomatic) radiation pneumonitis in 89% of patients. Future studies should emphasize selecting early-stage cancer patients receiving TRT as part of multimodal care (i.e., increased likelihood of long-term survival), determining the volume and dose of radiation to the heart, and in turn, the incidence of RIPE and delayed cardiovascular complications such as accelerated pulmonary hypertension or congestive heart failure. This will be a next step towards examining RIPEs in the context of multiple organ injury.

Statistical Analyses (performed by):

• Masooma Aqeel, MD (details above)

• Daniel Eastwood, MS (eastwood@mcw.edu), Dept. of Biostatistics, Health Research Center, 8701 Watertown Plank Rd, Milwaukee, WI 53226, United States. +1 (414) 955 4855

• Anjishnu Banerjee, PhD (abanerjee@mcw.edu), Dept. of Biostatistics, Health Research Center, 8701 Watertown Plank Rd, Milwaukee, WI 53226, United States +1 (414) 955 8358