During 1996–2012, we identified 8983 patients aged 15–39 years diagnosed with NHL in California. After exclusion of patients who died within two years after diagnosis (n = 2995), had unknown survival time (n = 94), those with unknown/invalid linkage to OSHPD data (n = 1049), and those with a SPM within 60 days after diagnosis or with an unknown date of SPM (n = 28), we identified 4817 AYA survivors. Of those, 4392 (91·2%) were HIV-uninfected and 425 (8·8%) were HIV-infected patients. The median follow-up time was 9·5 and 9·3 years (range 2·0–19·0) for HIV-uninfected and HIV-infected patients, respectively. Overall, 23·8% HIV-uninfected and 39·8% HIV-infected AYAs had ≥1 late effect(s).

Table I shows the baseline characteristics of the NHL survivors by HIV status. The majority of patients were male, of white race/ethnicity, diagnosed at age 30 or older, lived in lower nSES, and had DLBCL. HIV-infected vs. HIV-uninfected survivors had a higher proportion of late stage disease (III/IV) at diagnosis (47·8% vs. 37·3%) and were about twice as likely to be uninsured/publicly (48·0%) vs. privately (21·4%) insured. HIV-uninfected AYAs were over twofold as likely to receive HSCT than HIV-infected patients (13·3% vs. 5·4%). HIV-uninfected patients had a higher proportion of follicular lymphoma, NK/T cell lymphoid neoplasm and lymphoblastic leukaemia, and a lower proportion of DLBCL and Burkitt lymphoma, than HIV-infected patients. Between 1996–2000 and 2009–2012, we observed a decreased utilisation of HSCT (18·1% and 9·0%, P < 0·001) and radiation (33·8% and 21·9%, P < 0·001) among HIV-uninfected survivors. No significant differences in HSCT and radiation were observed among HIV-infected patients over time (data not shown in Tables).

Among HIV-uninfected AYAs, the most frequent diseases at 10 years were as follows: endocrine (CMI = 18·5%), cardiovascular (CMI = 11·7%), and respiratory (CMI = 5·0%), followed by SPM (CMI = 2·6%), renal and neurologic (CMI = 2·2%), liver/pancreatic (CMI = 2·0%), and AVN (CMI = 1·2%). The two most common SPMs among HIV-uninfected survivors were breast (19·2%) and skin melanoma cancers (12·3%), whereas among HIV-infected survivors, 54·5% of SPM was anorectal cancer (footnotes of Tables II and III, respectively). HSCT was associated with an increased risk of all late effects among HIV-uninfected survivors (Table II). HIV-infected survivors had a higher CMI of all late effects compared to HIV-uninfected patients (Fig 1).

When compared to other racial/ethnic groups, HIV-uninfected blacks and Hispanics had a higher incidence of cardiovascular, renal, endocrine and neurologic diseases, whereas HIV-infected blacks had a greater incidence of cardiovascular, renal, respiratory and neurologic diseases. HIV-uninfected patients with advanced stage disease (III/IV), who were uninsured/publicly insured, and those who lived in lower SES neighbourhoods, had a greater CMI of most late effects. Conversely, among HIV-infected patients, these factors were not significantly associated with the CMI of late effects (Tables II and III).

In multivariable models, among HIV-uninfected patients, public/no insurance (vs. private) was associated with greater risk of most late effects, except AVN. Likewise, receipt of HSCT was associated with a higher risk of most diseases, including AVN, cardiovascular, respiratory, renal and endocrine diseases. The risk of developing late effects varied by nSES, with AYAs who resided in lower (vs. higher) nSES at a higher risk for developing cardiovascular, respiratory and endocrine diseases. Hispanic and black race/ethnicity (vs. white) were associated with a higher risk of renal disease. Male survivors had a higher risk of renal disease, while females were at increased risk of SPM. Our analysis by NHL subtype revealed that lymphoblastic lymphoma was significantly associated with an increased risk of endocrine disorders and AVN (Table IV). In multivariable models limited to HIV-infected patients, we observed a higher risk of renal disease among non-white patients, without evident associations with other factors (Table V). We did not observe differences in late effects by calendar year of diagnosis, except for a higher risk of neurologic diseases among HIV-uninfected survivors in the later (vs. earlier) period of diagnosis (Table IV).

Advances in the management of NHL with risk-stratified and multimodal treatment have led to a significant survival improvement. However, NHL AYA survivors face a lifelong risk of developing late effects. The therapeutic toxicity after NHL has been demonstrated in several studies of childhood cancer survivors (Oeffinger et al., 2006; Bhakta et al., 2016; Bhakta et al., 2017), but just a few reports were specific for NHL (Haddy et al., 1998; Ehrhardt et al., 2017). To our knowledge, this is the first population-based study to investigate the incidence of late effects among AYAs with NHL, and also to explore the impact of sociodemographic and clinical factors on the occurrence of these conditions among HIV-uninfected and -infected patients. We found that AYAs who survived ≥2 years after NHL diagnosis had a substantial burden of late effects, particularly those with public or no insurance, HIV-infected patients, HSCT recipients, and survivors who lived in lower SES neighbourhoods.

The majority of NHL patients receive chemotherapy, and consequently late effects among AYAs are most likely secondary to the cumulative doses and intensities of the chemotherapeutics, mainly anthracycline and alkylating agents (Ehrhardt et al., 2019). Our findings of a higher incidence of cardiovascular and endocrine diseases are consistent with previous childhood studies (Haddy et al., 1998; Meacham et al., 2010; Bhakta et al., 2016; Bhakta et al., 2017; Ehrhardt et al., 2017). For example, a report from the Childhood Cancer Survivor Study showed that young survivors of childhood cancers were more likely to take medications for hypertension, dyslipidemia, and diabetes than their siblings (Meacham et al., 2010). Additionally, a more recent study at a single US institution showed that the four most prevalent late effects among children/adolescent NHL survivors were being overweight/obesity, increased fasting glucose, high total cholesterol, and hypertension (Ehrhardt et al., 2017). Another report from the St. Jude Lifetime Cohort Study (Bhakta et al., 2017) demonstrated that among patients treated for childhood cancer who survived ≥10 years and were aged ≥18 years, respiratory disease and SPM contributed to the overall cumulative burden and severity of chronic medical conditions, supporting our results.

We observed that the majority of SPMs among both HIV-uninfected and HIV-infected survivors were solid tumours. HIV-uninfected AYAs were more likely to develop breast and melanoma skin cancers, whereas over half of SPMs among HIV-infected survivors were anorectal cancers. A report has shown (Wang et al., 2018) that some childhood cancer survivors carry germline mutations in cancer predisposition genes, which are significantly associated with increased rates of solid tumours in irradiated survivors (breast cancer and sarcomas) and non-irradiated survivors (any SPM, breast cancer, non-melanoma skin cancer, and ≥2 histologically different SPMs), supporting the referral of AYA NHL patients for genetic counselling. Recent studies have suggested that chemotherapy with anthracycline and alkylating agents also increase the risk of SPM (Henderson et al., 2016; Teepen et al., 2017). Interestingly, we found a more pronounced increase of SPM in HIV-infected versus HIV-uninfected survivors. Multiple factors may have contributed to the increased incidence of SPM in HIV-infected survivors, including chronic antigenic stimulation, chronic inflammation, cytokine deregulation, and concomitant infections (Yarchoan & Uldrick, 2018). Additionally, HIV-infected patients are nearly twice as likely to smoke, and find it more difficult to quit smoking, than the general population (Mdodo et al., 2015), further increasing the risk of developing smoking-related cancers, as well as cardiovascular and respiratory diseases. Our findings emphasise the importance of AYA long-term surveillance to prevent (e.g. through sun protection and smoking prevention/cessation) and detect SPM early on (e.g. mammography for breast cancer screening, and physical examination and anoscopy/sigmoidoscopy for anorectal cancers).

Approximately 13% of HIV-uninfected AYAs received HSCT. We found that these patients were at greater risk of presenting with cardiovascular, respiratory, renal, endocrine and AVN than non-HSCT recipients. Although patients who received HSCT usually have a good quality of life, they have a lifelong risk of late health complications, which are more frequently seen than after conventional chemotherapy and/or radiation (Bhatia, 2011; Smith et al., 2013). These findings underline the importance of optimal survivorship care in this population.

Our analysis by NHL subtypes showed that among HIV-uninfected patients those with lymphoblastic lymphoma had the highest CMI of AVN (likely due to a high dose of steroids), whereas follicular lymphoma had a greater CMI of cardiovascular, respiratory and endocrine diseases. Patients with Burkitt lymphoma presented the lowest CMI of endocrine and cardiovascular diseases compared to other histological subtypes. We did not observe significant differences in the CMI of late effects by NHL subtype among HIV-infected survivors. These observations need to be interpreted in light of the differences in NHL subtypes between HIV-infected and HIV-uninfected patients, including a higher incidence of Burkitt lymphoma among HIV-infected, and follicular lymphoma among HIV-uninfected patients. While the former patients may have received more intensive multidrug therapy, the later may have received less intensive treatment or a ‘watch and wait’ approach. Furthermore, potential differences in treatment management over time such as reduction in the use of radiation, introduction of immunotherapy combined with chemotherapy for some types of NHL, and improvements in antiretroviral therapy may have influenced the occurrence of late effects we observed.

Uninsured/publicly insured HIV-uninfected AYAs had a higher risk of developing late effects than privately insured survivors. This finding suggests that they have less access to healthcare as well as suboptimal preventive survivorship care. A US report, which used data from the Behavioral Risk Factor Surveillance System, revealed that AYA cancer survivors may avoid healthcare due to financial barriers, precluding prevention and early detection of late adverse effects (Kirchhoff et al., 2012). The authors appropriately highlighted that although expansion of health insurance coverage for AYAs is important, this may not be sufficient to guarantee optimal care for these patients, unless cost-sharing and high out-of-pocket expenses are removed/reduced.

Similarly, nSES also influenced the occurrence of late effects among HIV-uninfected patients. AYAs who lived in lower SES regions had a higher risk of cardiovascular, respiratory and endocrine diseases. These findings are consistent with an earlier study on AYAs with Hodgkin lymphoma in California, which found that those living in lower SES neighbourhoods had a higher incidence of respiratory and endocrine diseases (Keegan et al., 2018). These data corroborate our colleagues’ interpretation (Kirchhoff et al., 2012; Kent et al., 2013; Keegan et al., 2018) that financial barriers can lead to dramatic consequences for the quality of life and survival of AYAs with cancer, as a result of delaying or foregoing much needed healthcare. Conversely, health insurance and nSES were not associated with the development of late effects among HIV-infected survivors. This may relate to a more advanced stage at diagnosis and/or higher risk NHL among these patients, requiring more intensive cancer therapies.

Adolescent and young adults of Hispanic and black race/ethnicity had an increased risk of renal disease in both HIV-uninfected and HIV-infected patients. Consistent with studies in North America (Lucas et al., 2008; Abraham et al., 2015), black HIV-infected patients had a substantial increased hazard of renal disease compared to HIV-uninfected patients. This may be partially explained by the potential nephrotoxic effects of antiretroviral therapy, HIV-associated nephropathy and HIV-immune complex renal disease (D’Agati & Appel, 1998). In addition, recent studies have demonstrated that APOL1 renal risk variants have been implicated in the increased risk of end-stage renal disease among black versus white HIV-infected survivors (Jotwani et al., 2017).

Strengths of the present study include a large population of AYAs diagnosed with NHL in California who survived ≥2 years after diagnosis, comprising a highly diverse racial/ethnic population. While some previous studies used patient-reported late effects without medical validation, we used data from all types of non-federal hospitals, allowing the generalisability of our findings. Additionally, we were able to investigate the late effects among HIV-infected survivors, a population which continues to grow and which endures a higher morbidity and mortality than HIV-uninfected individuals.

Adolescent and young adult NHL survivors had a wide spectrum of late effects, more specifically cardiovascular, respiratory, endocrine and secondary malignancies. We identified subgroups of patients with higher risk of diseases, such as uninsured/publicly insured AYAs, those who lived in lower SES neighbourhoods, HSCT recipients and HIV-infected patients. A deepened understanding of the late adverse effects of NHL treatment can facilitate the development of an effective survivorship plan, focusing on prevention, early detection and treatment of these conditions. Expansion of healthcare coverage and reduction of financial barriers may contribute to better long-term care of AYAs with NHL, ultimately improving the survival and quality of life of this population.