Early evidence of pulmonary dysfunction in survivors of childhood Hodgkin lymphoma

Abstract

Survivors of childhood Hodgkin lymphoma (HL) are at risk for pulmonary late effects, but whether survivors also experience pulmonary dysfunction early off therapy is not well understood. We determined the incidence of pulmonary dysfunction in children/adolescents with HL at entry into survivorship, as well as risk factors related to this outcome. Survivors in clinical remission and with a pulmonary function test (PFT) obtained 2–6 years off therapy were included. Seventy-five of 118 subjects met eligibility criteria (mean age at diagnosis: 13 years, mean time off therapy: 40 months). Survivors of HL had a higher than expected incidence of pulmonary dysfunction at entry into survivorship (40/75 [53%] had an abnormal DLCO and/or a restrictive or obstructive impairment). Evidence for diffusion impairment was associated with female sex (odds ratio [OR] = 3.19, p = .04). Longitudinal follow-up studies are needed to determine if early evidence of pulmonary dysfunction predicts risk for later onset pulmonary outcomes.

Introduction

Advances in childhood Hodgkin lymphoma (HL) therapy have led to relatively high cure rates in the modern treatment era; however, this success is tempered by therapy-related sequelae. For example, chemotherapy and radiation therapy (RT) used to treat high-risk HL lead to pulmonary complications in approximately 30% of survivors by the age of 45 years and are a leading cause of death in this population [1,2]. The prevalence of pulmonary late effects in survivors of HL supports routine surveillance with pulmonary function tests (PFTs) at entry into long-term follow-up care, with ongoing surveillance recommended if the initial study is abnormal [3,4].

The majority of studies describing pulmonary complications in survivors of childhood cancer have been conducted in adult populations comprising survivors of various primary diagnoses that were treated with regimens from earlier treatment eras. These studies describe late-onset pulmonary outcomes of chemotherapy and radiation that was administered in a variety of disease contexts, and may be less relevant to outcomes of contemporary therapy [2,5–7]. Fewer studies have investigated the incidence of pulmonary dysfunction early off therapy, i.e. at the time of entry into long-term follow-up care, and again include survivors of various primary diagnoses [8–11]. The varying degree to which survivors with at-risk exposures develop pulmonary complications suggests that host factors may contribute to this risk. An understanding of factors contributing to risk for pulmonary dysfunction at the entry to survivorship may lead to improved identification of at-risk populations that could benefit from more frequent surveillance and early referral to pulmonology. The objective of this study is to determine the incidence of pulmonary dysfunction in survivors of childhood HL at entry into long-term follow-up care (2–6 years off therapy), and to identify host and treatment factors that are associated with this outcome.

Methods

Study population

The study population included survivors of classical HL treated between 1991 and 2016 at Texas Children’s Hospital (TCH) and diagnosed at ≤22 years of age (Figure 1). For inclusion in this study, survivors must 1) have been treated with agents known to confer risk for adverse pulmonary outcomes (i.e. bleomycin and/or RT), 2) be in remission and have had no history of relapse or progressive disease, and 3) have had at least one PFT obtained between 2 and 6 years off therapy that included measures of spirometry with or without diffusion capacity of the lung for carbon monoxide (DLCO). Demographic and exposure data were obtained on all evaluable survivors, and the volume of lung tissue exposed to RT (mean lung dose [MLD]) and the percent of lung exposed to ≥20 Gy [V20]) were obtained when available. This research was conducted under the Baylor College of Medicine Institutional Review Board-approved protocols and in accordance with the Declaration of Helsinki.

Figure 1.

Survivors of Hodgkin lymphoma meeting eligibility criteria. Abbreviations: HL: Hodgkin lymphoma; RT: radiation therapy; PFT: pulmonary function test; DLcoC/VAadj: diffusion capacity of the lung: corrected for hemoglobin and alveolar volume.

Assessment of pulmonary function

PFTs were obtained as part of routine clinical care in the TCH Pulmonary Diagnostic Laboratory, and in accordance with American Thoracic Society standards. Equipment is calibrated daily, and includes a Jaeger MasterScreen Spirometer (Vyaire Medical Inc, Würzburg, Germany) and a single-breath analyzer for spirometry and determination of DLCO. For this study, DLCO was corrected for hemoglobin and adjusted for alveolar volume (DLcoC/VAadj), adhering to recommended consensus guidelines for optimal DLCO assessment [12]. Measured values were compared against normal predicted values for the patient’s age, sex, height, and race, when applicable, and resulted as percent predicted values. Our analyses were based on the first PFT that was completed for each survivor and that was obtained at least 2 years and up to 6 years off therapy (index PFT), coinciding with the timing of referral to the TCH Long-Term Survivor Clinic. Abstracted PFT data were reviewed and evaluated by a single pediatric pulmonologist who was blinded to survivor outcomes.

Abnormal PFTs were characterized as indicative of obstructive or restrictive impairment, diffusion impairment, or a combined obstructive or restrictive impairment plus diffusion impairment. Forced vital capacity (FVC) and forced expiratory volume in one second (FEV1) were considered abnormal if <80% predicted. Obstructive impairment was defined as having an abnormal FEV1 and/or a FEV1/FVC ratio of <0.8, while restrictive impairment was defined as having an abnormal FEV1 and FVC with an FEV1/FVC ratio of ≥0.8, as previously defined [13]. Diffusion impairment was defined using the Common Terminology Criteria for Adverse Events version 3.0 (CTCAE v3.0), the latest version to include specified thresholds for DLCO in the grading criteria. For this study, diffusion impairment was defined as CTCAE grade 2 or higher, corresponding with a DLcoC/VAadj that was <75% of predicted [14], as has previously been applied in childhood cancer survivor populations [6].

Clinical assessment

Age at diagnosis, sex, race, ethnicity, date of treatment start and completion, disease stage, risk group (low/intermediate-risk vs. high-risk), cumulative bleomycin dose, presence/absence of thoracic surgery, and clinical symptoms assessed near the time of the index PFT were abstracted from the medical record. RT data were collected as a yes/no variable (for both ‘any’ RT and ‘lung’ RT), as prescribed doses (for both ‘any’ RT and ‘lung’ RT), and, when available, as dosimetric variables (MLD and V20).

Radiographic assessment

Off-therapy surveillance computed tomography (CT) scans obtained within 24 months of the index PFT (n = 41; mean = 12.1 months from PFT, range: 24 months before to 24 months after index PFT) were re-reviewed for evidence of interstitial lung disease by a single pediatric radiologist reviewer who was blinded to survivor outcomes. Those without signs of interstitial lung disease were given a score of 0, and those with signs of mild, moderate, or severe disease were given a score of 1, 2, or 3, respectively, using methods that have been previously described [15].

Statistical analysis

Descriptive statistics, including means and standard deviations for continuous variables and counts, and percent of the total for categorical variables, were calculated for demographic and treatment characteristics. Associations between treatment/demographic characteristics and pulmonary outcomes were modeled using multivariable linear (DLcoC/VAadj, FEV1, FVC, FEV1/FVC) or logistic (diffusion, obstructive, restrictive impairment) regression. Covariates were included in multivariable regression models and were selected using a backward stepwise procedure, whereby a parsimonious model included only statistically significant (p < .05) or clinically relevant (e.g. sex, age) factors. Statistical analyses were conducted in Stata 14.0 (Stata Corp, College Station, TX) at a 2-sided p < .05 significance level.

Results

Cohort characteristics

There were 118 patients diagnosed with HL within the timeframe of 1991 and 2016. Of these, 10 were excluded due to relapsed or progressive disease, and one received neither bleomycin nor RT as part of their treatment. Of the remaining 107, thirty-two were excluded due to failure to obtain a PFT between 2 and 6 years off therapy (Figure 1). Those excluded did not differ by age at diagnosis, race/ethnicity, disease stage, or receipt of RT, though more males were excluded than females (Supplemental Table 1). Of the remaining 75 survivors, all had PFTs that included spirometry and 63 had both spirometry and DLcoC/VAadj. Absence of DLcoC/VAadj (n = 12) was due to (a) the lack of recent hemoglobin value permitting adjustment (n = 1), (b) inability of patient to comply with assessment (n = 6), (c) incorrect order placed by provider (n = 3), or (d) reasons that were undocumented (n = 2). Age range, time from therapy completion, demographics, and relevant disease and treatment features were compared for survivors with and without pulmonary dysfunction, shown in Tables 1 and 2.

Table 1.

Demographic and clinical features of subjects with obstructive or restrictive abnormalities compared to those without pulmonary function defects, n = 75.

	Normal spirometry	Obstructive abnormalities		Restrictive abnormalities
	n = 62	n = 10	p-value	n = 3	p-value
Mean age at diagnosis, year (SD)	13.1 (3.8)	12.8 (3.3)	.85	14.7 (2.1)	.47
Age at diagnosis, n (%)			.91		.78
<10 years	11 (17.7)	2 (20.0)		0 (0.0)
10–14 years	24 (38.7)	3 (30.0)		2 (66.7)
≥15 years	27 (43.6)	5 (50.0)		1 (33.3)
Race/ethnicity, n (%)			.71		.60
Non-Hispanic White	27 (43.6)	6 (60.0)		0 (0.0)
Non-Hispanic Black	7 (11.3)	1 (10.0)		0 (0.0)
Hispanic	28 (45.2)	3 (30.0)		2 (100.0)
Sex, n (%)			.17		.99
Male	22 (35.5)	6 (60.0)		1 (33.3)
Female	40 (64.5)	4 (40.0)		2 (66.7)
Treatment risk group, n (%)			.99		.55
LR/IR	48 (77.4)	8 (80.0)		2 (66.7)
HR	14 (22.6)	2 (20.0)		1 (33.3)
Disease stage, n (%)			.29		.99
Stage 1	1 (1.6)	0 (0.0)		0 (0.0)
Stage 2	33 (52.2)	6 (60.0)		2 (66.7)
Stage 3	8 (12.9)	3 (30.0)		0 (0.0)
Stage 4	20 (32.3)	1 (10.0)		1 (33.3)
EBV, n (%)			.42		.99
None	39 (69.6)	4 (50.0)		2 (66.7)
Any	17 (30.4)	4 (50.0)		1 (33.3)
B-symptoms, n (%)			.74		.55
None	39 (62.9)	7 (70.0)		1 (33.3)
Any	23 (37.1)	3 (30.0)		2 (66.7)
Smoking statue, n (%)			.57		.99
Never	47 (92.2)	8 (88.9)		3 (100.0)
Ever	4 (7.8)	1 (11.1)		0 (0.0)
Thoracic surgery, n (%)			.59		.37
None	54 (87.1)	10 (100.0)		2 (66.7)
Any	8 (12.9)	0 (0.0)		1 (33.3)
RT, n (%)			.69		.88
None	15 (24.6)	3 (33.3)		0 (0.0)
Any	46 (75.4)	6 (66.7)		3 (100.0)
Lung RT, n (%)			.14		.55
None	22 (36.1)	6 (66.7)		0 (0.0)
Any	39 (63.9)	3 (33.3)		3 (100.0)
Prescribed lung RT dose, n (%)			.22		.61
None	22 (36.1)	6 (66.7)		0 (0.0)
21 Gy	36 (59.0)	3 (33.3)		3 (100.0)
>21 Gy	3 (4.9)	0 (0.0)		0 (0.0)
Percent of lung receiving ≥20 Gy (V20), n (%)			.02		1
< 20 %	29	9		1
≥ 20 %	23	0		1

Significant values are in bold (p < .05). Abbreviations: LR/IR: low-risk/intermediate-risk; HR: high-risk; EBV: Epstein-Barr Virus; RT: radiation therapy; Gy: Gray; MLD: mean lung dose.

Table 2.

Demographic and clinical features of subjects with diffusion abnormalities are compared to those without these abnormalities, n = 63.

	Normal diffusion	Abnormal diffusion
	n = 32	n = 31	p-value
Mean age at diagnosis, year (SD)	12.8 (4.0)	14.3 (2.7)	.09
Age at diagnosis, n (%)			.62
<10 years	6 (18.8)	3 (9.7)
10–14 years	12 (37.5)	11 (35.5)
≥15 years	14 (43.7)	17 (54.8)
Race/Ethnicity, n (%)			.29
Non-Hispanic White	11 (35.5)	17 (54.8)
Non-Hispanic Black	4 (12.9)	3 (9.7)
Hispanic	16 (51.6)	11 (35.5)
Sex, n (%)			.04
Male	17 (53.1)	8 (25.8)
Female	15 (46.9)	23 (74.2)
Treatment risk group, n (%)			.39
LR/IR	26 (81.2)	22 (71.0)
HR	6 (18.8)	9 (29.0)
Disease stage, n (%)			.06
Stage 1	0 (0.0)	1 (3.2)
Stage 2	16 (50.0)	18 (58.1)
Stage 3	8 (25.0)	1 (3.2)
Stage 4	8 (25.0)	11 (35.5)
EBV, n (%)			.39
None	18 (62.1)	22 (75.9)
Any	11 (37.9)	7 (24.1)
B-symptoms, n (%)			.20
None	22 (68.8)	16 (51.6)
Any	10 (32.2)	15 (48.4)
Smoking statue, n (%)			.99
Never	24 (92.3)	24 (88.9)
Ever	2 (7.7)	3 (11.1)
Thoracic surgery, n (%)			.15
None	30 (93.7)	25 (80.7)
Any	2 (6.3)	6 (19.4)
RT, n (%)			.13
None	10 (31.3)	4 (13.3)
Any	22 (68.7)	26 (86.7)
Lung RT, n (%)			.30
None	14 (43.8)	9 (30.0)
Any	18 (56.2)	21 (70.0)
Prescribed lung RT dose, n (%)			.15
None	14 (43.8)	9 (30.0)
21 Gy	16 (50.0)	21 (70.0)
>21 Gy	2 (6.25)	0 (0.0)
Percent of lung receiving ≥20 Gy (V20), n (%)			.03
<20 %	21	12
≥20 %	7	14

Significant values are in bold (p < .05). Abbreviations: LR/IR: low-risk/intermediate-risk; HR: high-risk; EBV: Epstein-Barr Virus; RT: radiation therapy; Gy: Gray; MLD: mean lung dose.

The mean time off therapy was 40 months (range 24–71 months). Seventy-two survivors received adriamycin, bleomycin, vincristine, etoposide, prednisone, cyclophosphamide (ABVE-PC)-based regimens, one received AVPC, one received cyclophosphamide, vincristine, procarbazine, prednisone, doxorubicin, bleomycin, vinblastine (COPP/ABV), and one received bleomycin, etoposide, doxorubicin, cyclophosphamide, vincristine, procarbazine, prednisone (BEACOPP) followed by COPP/ABV. Regarding potentially pulmonary-toxic therapies, 73 survivors were treated with bleomycin (mean cumulative dose = 62.7 IU/m², range: 45–90 IU/m²) ± RT, and 2 received RT only (Figure 1). Of the 57 subjects who received RT, 45 were treated with lung RT (e.g. delivered to the chest or mediastinum), of which 44 had available dosimetric data: mean V20 14.3%, MLD 6.1 Gray (Gy). Clinical symptoms were assessed nearest to the time of the index PFT (mean 8 days, range: 0–100 days). While no subjects reported chronic cough, need for oxygen, or frequent infections, five reported shortness of breath (n = 3) and/or chest pain (n = 3). The corresponding index PFT revealed a diffusion defect in one subject, diffusion and obstructive defects in one subject, and normal PFTs in the other three subjects.

At time of entry to long-term follow-up care, 17% of survivors had either an obstructive or restrictive impairment on spirometry (13/75), and 49% had diffusion impairment (31/63)

Ten out of 75 survivors (13.3%) had an obstructive impairment, and 3/75 had a restrictive impairment (4.0%). None of the 10 survivors with an obstructive impairment had a diagnosis of asthma, and only one had a history of tobacco use. Of the 63 survivors with PFTs that included DLcoC/VAadj (mean 74% of predicted, range 52–105% of predicted), 27 (42.9%) had an isolated diffusion impairment and 4 (6.3%) had a combined defect that included both diffusion impairment and either obstructive or restrictive impairment. All cases with diffusions impairment met criteria for Grade 2 by CTCAE v3.0 (DLcoC/VAadj ≥50 and <75% of predicted) (Figure 2). Forty-one survivors had an off-therapy CT scan that was within two years of the index PFT, and 37 of these survivors had PFTs that included DLcoC/VAadj. Twenty-five of these 37 survivors (68%) had an abnormal DLcoC/VAadj. However, only two with an abnormal DLcoC/VAadj had evidence of interstitial lung disease (mild or Grade 1), therefore it is unlikely that the high incidence of diffusion abnormalities was due to early-onset pulmonary fibrosis, a known late effect associated with bleomycin exposure.

Figure 2.

Distribution of type of PFT abnormality at entry into long-term follow-up care among survivors with comprehensive PFTs that included DLcoC/VAadj: n = 63. Abbreviations: PFT: pulmonary function test; DLcoC/VAadj: diffusion capacity of the lung: corrected for hemoglobin and alveolar volume.

Females had higher odds of diffusion impairment than males at time of entry to long-term follow-up care

In univariate analyses, female sex was significantly associated with diffusion impairment (p = .04). After multivariable analysis, the relationship between female sex and diffusion impairment remained significant (odds ratio [OR] = 3.19, 95% confidence interval [CI]: 1.04, 9.77, p = .04) (Table 3). Although V20 < 20% was associated with obstructive impairment (Table 1), and V20 ≥ 20% with diffusion impairment in univariate analysis (p = .03) (Table 2), these associations were no longer statistically significant after accounting for potential confounding variables in multivariable analysis.

Table 3.

Association between clinical/demographic factors and pulmonary abnormalities in multivariable analysis adjusting for variables included in the table.

	Abnormal diffusion		Obstructive abnormalities		Restrictive abnormalitiesa
	OR (95% CI)	p-value	OR (95% CI)	p-value	OR (95% CI)	p-value
Gender		.04		.07		.99
Male	1.00		1.00		1.00
Female	3.19 (1.04–9.77)		0.23 (0.05–1.12)		0.99 (0.07–13.61)
Age at diagnosis		.88		.46		.56
<10 years	1.00		1.00		1.00
10–14 years	1.30 (0.21–8.25)		0.94 (0.10–8.83)		2.24 (0.15–32.95)
≥15 years	1.55 (0.25–9.45)		2.56 (0.34–19.18)		––
Lung RT		.57		.13		–
None	1.00		1.00		1.00
Any	1.43 (0.42–4.91)		0.29 (0.06–1.43)		––
Treatment risk group		.27		.57		.89
LR/IR	1.00		1.00		1.00
HR	2.09 (0.57–7.66)		0.51 (0.05–5.21)		0.83 (0.06–11.68)

^aAll cases with restrictive impairment received XRT. Significant values are in bold (p < .05). Abbreviations: RT: radiation therapy; LR/IR: low-risk/intermediate-risk; HR: high-risk.

Discussion

Childhood cancer survivors are at increased risk for pulmonary complications after cancer therapy [2]. Risk and response-adapted therapy for HL limits exposure to specific chemotherapeutic agents and RT for those at lowest risk for relapse [16,17]. While late-onset pulmonary outcomes are well-characterized in at-risk survivor populations, pulmonary dysfunction at entry to long-term follow-up care is less studied. Here, we provide key insights for this outcome in survivors early off therapy, observing a higher than expected incidence of diffusion abnormalities using established definitions, and confirming prior observations suggesting a higher risk for this outcome among females.

Among survivors assessed by spirometry (n = 75), the incidence of obstructive and restrictive impairments observed were comparable to prior observations in early off-therapy survivors of HL [18]. However, we observed a remarkably higher incidence of diffusion abnormalities early in survivorship. This incidence exceeded reports from most studies conducted approximate to this follow-up time point [18–20]. These differences may be related to our approach that adjusted DLCO for hemoglobin and alveolar volume, in addition to our inclusion of age-based normative values (% predicted). Such adjustments were not made consistently in the above-cited work, but there is consensus that this approach provides a more accurate representation of true DLCO [12,21]. Of note, that all survivors included in our study had a mild impairment in DLCO (Grade 2, ≥50 to < 75% predicted) may reflect a heightened sensitivity for this method in detecting diffusion impairment. Also of note, there were minor differences in the time frame of ascertaining off-therapy PFTs (2–6 years) in our study compared with prior works (ranging from 1–20 years), which may have contributed to differences in the incidence of diffusion-related outcomes observed. [18–20] These results support the longitudinal assessment of pulmonary function in survivor populations, e.g. with serial PFTs measured non-discriminately for all survivors.

Both pneumonitis and late pulmonary function abnormalities, such as pulmonary fibrosis, are known complications of lung RT [2,18,22]. Subacute, earlier onset pneumonitis may also occur in patients with recent history RT exposure, though this is a relatively rare complication unlikely to contribute to the high incidence of diffusion abnormalities observed in this cohort [22]. Efforts to reduce RT exposure include the use of targeted techniques, such as three-dimensional conformal therapy and intensity-modulated RT, which limit the volume of tissue irradiated [23], and the omission of RT in low-risk populations [24]. Dosimetric data, such as MLD and V20, are increasingly used to predict the effect of radiation on the lungs. MLD in excess of 20 Gy has been associated with increased risk for both radiation pneumonitis and fibrosis. [25,26] Although we observed no risk association with MLD, none of our subjects met previously-published thresholds for increased risk, and only two had an MLD in excess of 15 Gy. Likewise, associations observed with V20 did not persist after multivariable analysis.

Female sex was associated with risk for diffusion impairment, an observation previously described in a study of 5+ year survivors of various childhood cancers treated with lung RT [5]. Female sex confers risk for various treatment-related complications in cancer survivors, including cardiovascular disease, second malignancies, obesity, and osteonecrosis [27]. The mechanism underlying these sex-related disparities is not well understood but may be related to pharmacokinetics, body composition, or differential expression of multi-drug resistant genes [27,28].

The Children’s Oncology Group Long-Term Follow-Up Guidelines recommend that at-risk survivors have a pulmonary function assessment with PFTs beginning two years after therapy completion (http://www.survivorshipguidelines.org). These recommendations are further substantiated by our findings of a high incidence of Grade 2 diffusion abnormalities early off therapy. Approximately 30% of our cohort did not adhere to the published guidelines, i.e. despite meeting exposure-based criteria they were not evaluated by a PFT that included spirometry and DLCO within the recommended timeframe. A number of barriers may preclude an appropriate assessment of diffusion capacity, such as age less than 6 years, inability of patients to comply with test instructions, and lack of recent hemoglobin to permit adjusting DLCO. The results of this study support a revision of the current guidelines to stipulate inclusion of DLCO that is corrected for both hemoglobin and alveolar volume (DLcoC/VAadj) with each PFT assessment to provide an optimal, comprehensive assessment of pulmonary function in at-risk survivors. Providers should be aware that in order to adjust DLCO for hemoglobin, a hemoglobin level must be obtained proximal in time to the assessment. Our analysis relied on toxicity criteria definitions from 2003 as the latest set of criteria available that provides specified thresholds for abnormal DLCO. There is a need for establishing appropriate DLCO cutoffs (adjusted for hemoglobin and alveolar volume) to define diffusion impairment, particularly among survivor populations.

Strengths of this study are our focus on HL survivors treated according to contemporary risk and response-adapted stratification. However, there were also some limitations. First, because this was a retrospective study, PFTs were not obtained at the same time off therapy for all subjects. Also, 30% of the eligible survivors of HL had neither spirometry nor DLCO assessed during the time period of interest. A comparison of those included versus those excluded did not reveal any obvious selection bias; however, it is possible that survivors who were compliant with off-therapy PFTs were those who also experienced pulmonary symptoms during treatment, and were therefore more likely to have abnormal results. In addition, we acknowledge a potential for survival bias, as patients who did not survive to entry into long-term follow-up were excluded, and that results from a single institution may not be generalizable to all populations. Characterizing the incidence and type of pulmonary dysfunction observed in survivors recently off therapy, and factors associated with these abnormalities, provides a critical foundation for longitudinal follow-up studies seeking to define early indicators of risk for later onset complications.