There is now strong evidence supporting the benefits of exercise to mitigate numerous TRTs in a range of adult cancer cohorts [12, 13]. Despite this, evidence in AYAs remains limited, with respect to both the impact of exercise on TRTs as well as the collection and reporting of TRTs in this cohort. To our knowledge, the current study is the first RCT to investigate the impact of an exercise intervention on patient-reported TRTs as collected using conventional CTCAE reporting methods. Additionally, this study is the first to report the trajectory of toxicities throughout treatment in AYA cancer patients. The most important finding of the current study was that exercise reduces the severity of fatigue in AYAs undergoing treatment. Additionally, the longitudinal analysis has provided more nuanced insight into the symptom burden experienced by AYAs undergoing cancer treatment.

Initially, when adopting conventional the maximal worst grade approach, results revealed the CG experienced a significantly greater incidence of severe fatigue (≥ Grade 3) compared with the EG. This fatigue, categorised as being not relieved by rest and limiting self-care activities, likely interrupted their daily functioning and has the potential to impact their ability to meet developmental milestones [23]. Supporting previous evidence, this study demonstrated the benefits of exercise in reducing fatigue in AYAs undergoing treatment [23]. No other significant differences were identified between groups for any toxicities using this conventional analysis approach.

Regardless of age, diagnosis or specific treatment protocol, patients undergoing cancer treatment often experience a myriad of TRTs [24]. Standard toxicity reporting in clinical trial publications reports the maximal worst grade experienced at a single timepoint throughout treatment with a consensus that ≥ Grade 3 toxicities are of the greatest clinical relevance due to their potential life-threatening nature [25]. However, in recent years it has been recognised that the arbitrary cutoff of ≥ Grade 3 fails to recognise the development of toxicities over time as well as any persistent or chronic lower grade toxicities which often have deleterious impacts on patients’ abilities to tolerate treatment and their overall QOL [10]. Novel statistical analysis may better detect these lower grade but clearly troubling side effects [25]. Thanarajasingam et al. [21] developed an innovative approach to measuring toxicities that better reflects the “area under the curve” for low grade, but chronic, toxicities that may be clinically significant [21]. Therefore, through the adoption of longitudinal analysis using stacked bar charts and stream plots, unique insights into the possible interactions of exercise and TRTs were revealed.

Supporting previous research, the most common toxicities reported in the current cohort were fatigue, pain, nausea and mood disturbances [5, 6]. The ToxT approach was applied to these variables to provide insight into the patient experience of these toxicities over time [10]. Supporting the maximal worst grade results, the longitudinal analysis of fatigue revealed trends toward an increase in the incidence of severe fatigue throughout the weekly monitoring in the CG. Furthermore, there was a relative increase in mean fatigue over time in the CG with a comparable decrease evident in EG throughout the intervention period. These trends may also be apparent in the AUC results with a higher magnitude of fatigue reported in the CG, thus potentially limiting their ability to perform activities of daily living (ADLs). Collectively, this potential amelioration of fatigue as a result of exercise may contribute to AYAs being able to maintain their normal ADLs and remain engaged in treatment and psychosocial pursuits.

Chemotherapy-induced nausea has been reported to affect 18–81% of AYAs undergoing treatment [26]. Previous research has reported insufficient evidence to support the efficacy of exercise in mitigating nausea and vomiting [27]. However, Shim et al. (2019) reported smaller proportions of exercise group participants experiencing higher grade (Grades 3 and 4) nausea when compared with controls [28]. This was similarly reflected in our study with the EG not reporting any occurrences of Grade 3 nausea during the intervention compared to the CG who reported Grade 3 nausea for 30% of the intervention period. Given that poorly managed nausea can lead to dehydration, malnutrition and anorexia, the ability for the EG to avoid severe nausea may prevent them from experiencing such deleterious impacts and subsequently poorer outcomes [29]. Collectively however, longitudinal analysis revealed trends toward a reduction in nausea in both groups over time suggesting improved clinical management of nausea with effective anti-emetic regimes with each cycle of treatment. This effective management may in part be attributed to the rigorous recording of TRTs as result of enrolment in this study which was relayed to treating teams.

Previous research has reported more than 30% of AYAs undergoing treatment experience depression and anxiety symptoms [30]. The current results reflected this, with mood disturbances including anxiety and depression being the most cited TRTs. High-quality evidence now supports the use of exercise to reduce depression and anxiety in cancer cohorts [13]. In our longitudinal analysis, the EG appeared to experience less severe mood disturbances over the course of the intervention. AUC results potentially support this, with a lower magnitude of mood disturbances in the EG over time. Furthermore, the mean mood disturbances experienced by the EG appeared to reduce over time, with a comparative increase in the CG. This suggests that supervised exercise has the potential to help stabilise mood in AYAs undergoing cancer treatment.

Compared to the previous TRTs mentioned, our EG demonstrated trends toward more severe pain toxicity during the intervention period compared with the CG. This contrasts with a recent meta-analysis which reported significant favourable effects of exercise in mitigating pain, compared to a pooled control sample [31]. Given that pain toxicity in our study was a broad item, not capturing the specific cause of the pain, these results are difficult to interpret. We speculate that these results may be reflective of the delayed onset muscle soreness (DOMS) experienced in the EG during the intervention, which was avoided in the CG due to lack of additional structured activity. More rigorous pain symptomatology monitoring is warranted to gain further insight into the cause of pain and potential impact exercise may have.

We acknowledge a number of limitations in our study. Firstly, results may be affected by the inability to blind study participants and a further inability to prevent the CG from exercising during the intervention. Additionally, as this was a heterogenous cancer-diagnostic and treatment sample, the scheduling and toxicity profile variations of different treatment protocols are difficult to account for and this may have confounded results. Non-compliance to the study protocol may have also confounded results with four participants withdrawing prior to the 10-week assessment, subsequently reducing strength of the sample. Furthermore, the COVID-19 pandemic impacted study recruitment, exercise adherence and toxicity monitoring. Finally, this study was a secondary outcome of a larger RCT on which the sample size was powered to detect changes in Vo2peak (primary outcome). Given the large number of variables contributing to the toxicity monitoring, the current study was likely underpowered to detect significance in such a large number of outcome variables. Based on the trends observed in the current study, future research should be conducted and powered appropriately to detect changes in toxicities specifically.