Recommendations for surveillance of pulmonary dysfunction among childhood, adolescent, and young adult cancer survivors: a report from the International Late Effects of Childhood Cancer Guideline Harmonization Group

Methodology of International Late Effects of Childhood Cancer Guideline Harmonization Group (IGHG) and formulating key questions

Information about methods used to formulate IGHG recommendations was published previously (18). For our current recommendation, we organised a guideline panel of 33 members—representing COG, DCOG, UKCCLG, the Pan-European Network for Care of Survivors after Childhood and Adolescent Cancer (PanCare)—and pulmonary health and late effects experts from various medical specialties: paediatric oncology and haematology; paediatric and adult pulmonology; radiation oncology; epidemiology; and guideline experts (Appendix A) (19–21).

IGHG’s approach to formulating recommendations involves answering five key questions: 1) "Who needs surveillance?" 2) "What surveillance modality should be used?" 3) "At what frequency should surveillance be performed?" 4) “When should surveillance be initiated?” and 5) "What should be done when abnormalities are found?" Based on our preliminary literature search and a resulting absence of data, we performed the systematic literature search for only question 1) “Who needs surveillance?” and used expert opinions from paediatric and adult pulmonologists for questions 2–5. We did not use guidelines for surveillance of other pulmonary diseases, such as idiopathic interstitial pneumonitis or chronic obstructive pulmonary disease, due to different underlying pathophysiological mechanisms. Notably, guidelines for other diseases start with symptomatic patients diagnosed with specific pulmonary diseases, which differs from our population of interest—asymptomatic CAYA cancer survivors at risk of developing pulmonary problems.

Comparing existing guidelines and formulating clinical questions

First, we separately compared COG, DCOG, Scottish Intercollegiate Guidelines Network, and UKCCLG recommendations for each of the five key questions (15, 16, 19, 22). For the key question, “What should be done when abnormalities are found?” we also included vaccination and lifestyle factors. For the key question “Who needs surveillance?” we subsequently formulated 11 clinical questions and sub-questions to strengthen concordant recommendations and find consensus for discordant recommendations (Appendix B).

We used the PICO-framework to formulate clinical questions (23). Our population of interest included CAYA cancer survivors—defined by at least 50% of survivors diagnosed before age 30—who completed cancer treatment at least two years previously. We included potentially lung-toxic treatment modalities (selected chemotherapeutic agents, thoracic radiotherapy, thoracic surgery, and allogeneic HSCT) and tobacco exposure as interventions. We also included all chemotherapeutic agents mentioned in current LTFU guidelines as risk factors for pulmonary dysfunction (bleomycin, busulfan, and nitrosoureas [lomustine and carmustine]). Based on expert opinion, we additionally included treatment with cyclophosphamide, gemcitabine, and methotrexate. Comparators were considered during data extraction and differed between studies according to study design, such as non-exposed survivors, survivors exposed to lower chemotherapeutic doses, or community controls. Our outcome of interest was pulmonary dysfunction assessed by PFT. We focused on this single outcome because our preliminary literature search showed that other commonly reported clinical endpoints—in particular survivor-reported symptoms or clinician-reported diagnoses—had been assessed and reported heterogeneously and had the risk of subjectivity, such as different definitions of chronic cough or dyspnea (24–28). We categorised pulmonary dysfunction into four groups: obstruction (by FEV1, FEV1/FVC, MEF25–75%); restriction (by TLC, FVC); hyperinflation (by RV, RV/TLC); and diffusion capacity impairment (by DLCO). Although clinically relevant hyperinflation should be interpreted together with obstruction, we defined hyperinflation as a separate pulmonary outcome since it was defined as such in included studies.

Systematic literature search on "Who needs surveillance?"

We conducted our first systematic literature search in accordance with the preferred reporting items for systematic reviews and meta-analyses (PRISMA) guidelines restricted to PubMed and Ovid in November 2016 with updates in June 2019; December 2020; June 2022; and April 2023 (29). We developed our search strategy based on 11 clinical questions and five concepts: cancer diagnosis, population of CAYA cancer survivors, potential lung-toxic treatment modalities, pulmonary outcomes, and late effects (Appendix C). The inclusion criteria were given through the PICO-framework. We also excluded studies with fewer than 20 participants and studies only assessing prevalence of pulmonary dysfunction without measuring effect sizes of associations between exposures and pulmonary dysfunction (Appendix D).

Guideline panel members screened titles, abstracts, and full-texts. Two authors independently screened each study. Coordinators (MO, RK) resolved discrepancies. We extracted data from each study and entered it into evidence summary tables (Appendix E); separately performed risk of bias assessments for each study (Appendix F); and completed overall quality assessments of available evidence for each clinical question, according to the GRADE criteria (Appendix G) (18, 30). Each eligible study could contribute to answering more than one clinical question. For our overall conclusion of evidence, we summarised findings by type of pulmonary dysfunction (Appendix H).

Expert consensus on surveillance modality, start and frequency of screening, and procedures in case of pulmonary dysfunction

For questions about surveillance modality, start and frequency of screening, and procedures in cases of pulmonary dysfunction (key questions 2ℒ5), we held numerous meetings with paediatric and adult pulmonologists and guideline development experts. We formulated suggestions based on initial meetings, which we then discussed with guideline panel members until reaching consensus through an iterative approach with successive revisions and implementing suggestions from all panel members.

Translating evidence into recommendations

We assessed evidence and information we gathered from our systematic literature search and expert consensus within the evidence-to-decision framework—it weighs the impact of screening by estimating benefits and harms, resources and costs, impact on health inequities, acceptability, and feasibility. Through panel discussions, we achieved consensus for final recommendations, which we subsequently discussed with additional experts, including oncologists and survivors (Appendix A).

Comparison with existing guidelines

Comparing existing guidelines revealed relevant discordance (Appendix I). Since the Scottish guideline omitted recommendations for pulmonary dysfunction, we excluded it (22). The remaining three guidelines considered CAYA cancer survivors at risk for pulmonary dysfunction after exposure to bleomycin, busulfan, nitrosoureas, thoracic radiotherapy, or thoracic surgery. Practical details revealed discordances, including cut-off doses for chemotherapeutic agents or thoracic radiotherapy; radiation volume; age at treatment; and additional risk factors, such as renal dysfunction and pulmonary infection. All three guidelines recommended PFT, yet did not specify tests. We found no concordance for screening frequency and starting screening with five and ten years after diagnosis in the Dutch guideline; two years after completion of treatment in the COG guideline; and end of treatment in the UK guideline. Guidelines agreed about referring CAYA cancer survivors to pulmonologists in cases of pulmonary dysfunction; alerting anaesthetists about previous bleomycin treatment; advising survivors not to smoke; and considering pneumococcal and influenza immunisation (Appendix I).

General results from systematic literature search

Our systematic literature search identified 9284 studies. We assessed 704 full-texts for eligibility; 26 studies fulfilled inclusion criteria (Figure 1, Table 1) (1–3, 31–53). Reasons for excluding full-texts mainly included 1) assessing outcomes other than pulmonary function by PFT (n=186); 2) including non-CAYA cancer survivors (n=164); and 3) assessing outcomes fewer than two years after completing treatment (n=91). Most studies (n=12) included CAYA survivors of different cancer types; followed by studies on leukaemia (n=7), lymphoma (n=4), and brain tumors, neuroblastoma, and osteosarcoma with one study each. Appendix J contains key characteristics and our summary of evidence for each included study; Appendix K presents our evidence assessment summary and quality of data contributing to recommendations per clinical question. Quality of evidence was very low or low for most clinical questions (Table 2). We summarise primary reasons for downgrading the quality of evidence in Table 3 and provide more detail in Appendix K.

• Download figure
• Open in new tab

Figure 1: Flow diagram for selection of studies

(Studies could be included in more than one category)

Evidence on risk factors for pulmonary dysfunction among CAYA cancer survivors

We identified seven studies examining allogeneic HSCT as a risk factor for pulmonary dysfunction (31–37); 13 studies for thoracic radiotherapy (1–3, 31, 38, 40, 45–48, 50–52); five studies for thoracic surgery (1, 31, 38–40); between one and eight studies for selected chemotherapeutic agents (bleomycin, busulfan, nitrosoureas, cyclophosphamide, methotrexate); no studies for gemcitabine; and six studies for tobacco exposure (2, 40, 41, 43–45) (Table 2, Appendix K). For busulfan and nitrosoureas, only one study of very low quality was available; it provided insufficient evidence to decide whether these agents significantly impact pulmonary function (2). Four studies of low to very low quality of evidence assessed pulmonary function after cyclophosphamide-containing treatment (1, 42, 46, 47). Studies examining effects from active tobacco smoking showed contradictory results (2, 41, 43–45). No studies examined impact from passive tobacco smoking or cannabis use. Our clinical questions and sub-questions aimed to investigate impact from exposure versus non-exposure and also impact from different dose levels; age at exposure; chronic graft versus host disease; infections; and total body irradiation among individuals treated with HSCT (Appendix B). Between one and four studies examined impact from different cyclophosphamide, methotrexate, and bleomycin doses; yet no studies were available for different doses of nitrosoureas and busulfan. Seven studies examined the impact of age at treatment with HSCT and radiotherapy.

Overall, we identified several potential sources of bias and methodological issues in most studies (Table 3). Study design was retrospective in more than half (n=14), increasing risks of bias and non-standardised measurements. Half of the studies did not describe original population sizes from where they selected participants at risk. This makes it uncertain if results are internally and externally representative and can be extrapolated for the entire population of CAYA cancer survivors. Only half of the studies described how they performed pulmonary function testing, such as by implementing the joint European Respiratory Society (ERS)-American Thoracic Society (ATS) recommendations. Even though most studies (74%) reported reference values used to standardize CAYA cancer survivor PFT results, studies used 22 different sources of reference values—with up to ten different sources in one study (47). Only two studies used internationally recommended all-age reference values from the Global Lung Initiative (GLI) (3, 52). Cut-off value definitions, such as for restrictive disease, were inconsistent, which made PFT results difficult to interpret and impossible to compare between studies and age groups. Additionally, findings from different studies on the same exposure were frequently inconsistent or contradictory, such as with bleomycin (31, 38, 39). Inconsistent and contradictory aspects precluded analysing quantitatively and interpreting findings, such as a meta-analysis; prevented formulation of recommendations for specific pulmonary function abnormalities, such as obstructive, restrictive, and diffusion impairment; confounded analysis of effects of specific chemotherapeutic agents on pulmonary function; or prevented definition of cut-off doses for chemotherapeutics or radiotherapy. However, studies we examined provided some evidence CAYA cancer survivors treated with allogeneic HSCT, thoracic radiotherapy, and thoracic surgery are at risk for pulmonary dysfunction as measured by PFT.

Translating evidence and expert consensus into recommendations

Gaps in knowledge and future directions for research

With currently available evidence, we only answered a few of our original clinical questions and often to only a limited extent, such as any exposure to radiotherapy without differentiation for doses or volumes. To improve evidence on pulmonary dysfunction among CAYA cancer survivors, we outlined gaps in knowledge and methodological approaches for future research (Table 5).

We identified existing knowledge gaps for dose-response relationships of all studied exposures; newer chemotherapeutic or immunotherapeutic agents; other medical conditions, such as pulmonary complications during treatment, co-morbidities; impact of treatments on physiological processes affecting pulmonary function, such as lung growth and physiologic ageing; and also approaches for early assessment of pulmonary dysfunction and effects from preventive or curative interventions.

Most studies focused on well-established risk factors such as HSCT (31–37), thoracic radiotherapy (1–3, 31, 38, 40, 45–48, 50–52), and thoracic surgery (1, 31, 38–40). For chemotherapeutic agents—even those with previously reported pulmonary toxicity, such as bleomycin— we found clinical evidence for CAYA cancer survivors insufficient. Future studies evaluating other classical chemotherapeutic agents, targeted or immunotherapeutic agents, or pharmacovigilance data might identify new aspects of pulmonary dysfunction among CAYA cancer survivors. It is similar for radiotherapy, including a lack of data comparing photon and proton therapy where toxicity of protons might be lower from smaller irradiated volume compared with photons, possibly resulting in less lung-toxicity. For all exposures, we lack knowledge on how they interact with each other or how age at treatment or additional medical conditions modify impact from exposures; we also have little information about dose-response relationships.

Peak lung function attained in early adulthood and the trajectory of lung function decline with ageing are important for lung health across the life span. The impact of cancer itself, pulmonary complications during treatment, pulmonary co-morbidities, such as asthma, or impaired somatic growth, such as scoliosis, on peak attained lung function has not been examined. No studies examined whether CAYA cancer survivors start at a lower peak attained lung function or whether physiological ageing and decline in pulmonary function is faster and steeper than among the general population. Frailty and accelerated ageing were previously described for childhood cancer survivors (58–60). The definition of frailty is met when fulfilling three or more of five criteria: reduced lean muscle mass, weakness, slow walking speed, low energy expenditure, and fatigue (61). Ness et al. showed that components of frailty—reduced strength, walking speed, and increased fatigue—were as frequent among childhood cancer survivors from the St. Jude Lifetime Cohort at a median age of 33 years as among people age 65 years and older in the general population (62). By calculating the deficit accumulation index score, Williams et al. showed childhood cancer survivors acquire more damage and disease than community controls (60). Both studies suggest accelerated ageing among childhood cancer survivors. Factors contributing to accelerated ageing and frailty include more rapid cellular senescence, telomere length reduction, epigenetic modifications, somatic mutations, and mitochondrial DNA damage (58). These factors may also affect lung growth and function among children and adolescents or result in faster pulmonary ageing, but their potential impact among CAYA cancer survivors is unknown.

Another gap in knowledge concerns measuring early stages of pulmonary dysfunction. Prior studies primarily utilized spirometry, body plethysmography, or DLCO measurement. More sensitive tests, such as multiple breath washout tests, may identify pulmonary disease earlier and eventually contribute to better understanding of pulmonary dysfunction development among CAYA cancer survivors. Parisi et al. and Schindera et al. investigated pulmonary function of childhood cancer survivors using multiple breath washout tests (63, 64). Parisi et al. investigated 57 survivors with median follow-up time of 6.2 years from end of treatment; they did not show differences in ventilation homogeneity compared with controls (64). The 46 survivors evaluated by Schindera et al. were median 20 years from cancer diagnosis (63). Survivors defined as high risk (bleomycin, busulfan, nitrosoureas, HSCT, thoracic radiotherapy, or surgery) tended to have more ventilation inhomogeneity than those at standard risk (other cancer therapies), yet not significantly (63). In both studies, more survivors had abnormal washout tests than abnormal spirometry.

Available data made it impossible to reach conclusions about the longitudinal course of pulmonary function as survivors progress through childhood, puberty, and adulthood in their growth and development followed by a trajectory of ageing. Eight studies with repeated PFT results suffered from attrition bias, small sample sizes, and included sub-groups of CAYA cancer survivors, such as HSCT (2, 32, 33, 35, 37, 47, 52, 53). Ideally with baseline PFT before starting treatment, longitudinal data— ascertained at regular intervals from diagnosis—will be available improve knowledge about the onset of pulmonary dysfunction and its evolution.

PFT provides one way of assessing pulmonary health. Clinical symptoms or imaging are other possible modalities. Clinical symptoms lack objectivity and vary with age, which limits precise measurement. In addition, questions about clinical symptoms are worded differently between studies, which makes comparisons difficult. Louie et al validated selected self-reported complications from HSCT survivors (65). No data exist for other CAYA cancer survivor populations or for other questions about pulmonary dysfunction.

We suggest future studies take weaknesses and methodological problems we identified from the literature we reviewed into account and avoid them whenever possible (Table 5). Collaboration between paediatric oncologists and pulmonologists helps avoid shortcomings when conducting PFT and reporting results. Collaborative studies with harmonised protocols could maximise statistical power with larger numbers of CAYA cancer survivors and allow stratifying analyses into sub-groups defined by therapeutic exposures, age at treatment, cumulative doses, or genotype. Prospective rather than retrospective studies allow for standardising assessments, such as PFT at predefined time points, and minimise selection and attrition biases. Assessing patient-reported outcomes, such as symptoms, functional limitations, and quality of life, together with PFT, helps determine the clinical significance of findings and their impact on lives of patients and their families.

To obtain accurate measurements, PFT must be performed to high standards, by trained personnel explicitly applying published guidelines and standards, including ERS/ATS guidelines. Reporting, interpreting, and applying results in clinical practice are equally important. It is essential that future studies use GLI reference equations to standardise PFT results and make them comparable between age groups and regions (66, 67). Binary cut-offs—describing results as either normal or abnormal— reduce statistical power and introduce interpretations based on pre-defined threshold values. Since cut-off values differ between studies, such dichotomisation hampers comparisons of results, leading to conflicting and potentially misleading proportions of CAYA cancer survivors with pulmonary dysfunction. Reporting results as raw data and z-scores based on internationally agreed, age-adjusted reference values is preferred and allows comparing and pooling data.

We suggest studies investigating pulmonary function among CAYA cancer survivors be conducted in the knowledge that at present no curative treatments exist for suspected progressive inflammatory and fibrotic changes underlying pulmonary dysfunction. Therefore, we advise careful study of benefits and harms from repeat testing. However, awareness of impaired pulmonary function possibly leads to earlier treatment of bacterial infections with antibiotics, especially because excess pulmonary mortality and hospitalisations among CAYA cancer survivors are mainly from infection (68, 69). The U.S. Food and Drug Administration and European Medicines Agency approved two anti-fibrotic drugs—pirfenidone and nintedanib—for the treatment of idiopathic pulmonary fibrosis; patients with progressive-fibrosing unclassifiable interstitial lung disease possibly also benefit (70). Therefore, in the future the possible benefit from anti-fibrotic drugs could be an area for investigation among CAYA cancer survivors.

Strengths and limitations

Strengths of our recommendation are multidisciplinary and international collaboration, which included perspectives from paediatric and adult specialists in CAYA cancer care and survivors; broad inclusion criteria; our thorough review process paired with in-depth quality assessment of included studies; and resulting evidence. Limitations mainly reflect lack of available evidence: studies with small sample sizes, heterogeneous PFT result reporting, use of different reference values; and scarce longitudinal data.