Self‐reported cognitive outcomes among adolescent and young adult patients with noncentral nervous system cancers

Abstract Objective Cancer‐related cognitive impairment (CRCI) among adolescent and young adult (AYA) cancer patients with noncentral nervous system (CNS) cancers has not been well studied. In this study, we aimed to describe CRCI‐associated trends and characteristics among AYA cancer patients. Methods In a longitudinal cohort of AYA cancer patients without CNS disease, CRCI was evaluated over 1 year using the Functional Assessment of Cancer Therapy‐Cognitive Function Instrument, a self‐reported cognitive outcome measure. CRCI prevalence was quantified using the previously established minimal clinically important difference. CRCI‐associated longitudinal trends and factors were evaluated with mixed‐effects model analysis. Results Ninety‐one patients (mean age = 28.4 ± 6.7 years) were included. Approximately one‐third (34.1%) experienced CRCI at least once during the study follow‐up. Female gender (P = .02), Indian ethnicity (P < .01), current smokers (P < .01), anxiety/depressive symptoms (P < .01) and fatigue (P < .01) were found to be associated with poorer cognitive function among AYAs. Conclusions Although AYA cancer patients were relatively young and without CNS disease involvement, a significant proportion of them experienced clinically important decline in cognitive function. With improved understanding of this subject, effective strategies can be formulated to promote awareness of CRCI and mitigate its negative effects among AYA cancer patients.

cancer patients. However, the mean ages of the study subjects in these studies are generally much higher than the adolescent and young adult (AYA) age range, 2,3 which encompasses cancer patients between the ages of 15 and 39 years old, as defined by the Institute of Medicine. 4 Even among cancers that are common within the AYA age range, such as Hodgkin's lymphoma 5 and testicular cancer, 6 representation of AYA patients is often low, and subgroup analysis of AYA patients has not been presented in the literature. Age-specific CRCI studies have also been limited thus far to survivors of childhood cancers 7 and the elderly, 8 indicating a lack of robust studies quantifying the prevalence and characterizing CRCI trends in AYA patients diagnosed with malignancies.
Preliminary data suggest that AYAs are encumbered by CRCI at work or in school. The AYA Health Outcomes and Patient Experience study reports that up to 53% of working cancer survivors within this age range face problems with memory or paying attention at work or school. 9 In another cross-sectional study, AYAs who have undergone chemotherapy were also approximately 2 to 4 times more likely to experience difficulties with mental tasks while working than those who did not receive chemotherapy. 10 These issues are detrimental to AYAs, who are in a critical junction in life with many developmental milestones to achieve, including completing their education and embarking on a career. 11 Furthermore, the productivity loss associated with AYAs is disproportionately large since they serve as the backbone of the economy. 12 For these reasons, unaddressed CRCI can potentially lead to devastating consequences for both AYA cancer patients and society at large. It is therefore imperative to comprehensively evaluate and understand CRCI among AYA cancer patients.
We have previously conducted a longitudinal assessment of symptom burden and distress in a cohort of AYA cancer patients. 13 In this article, the self-reported cognitive function of patients with non-CNS malignancies will be reported. Furthermore, CRCI-associated characteristics will be identified. Our findings will be important to demonstrate if CRCI is prevalent in the AYA population.

| Study design
This was a secondary analysis of self-reported cognitive outcomes collected from a longitudinal cohort study at the National Cancer Centre Singapore. The study was conducted from September 2015 to October 2018. Ethics approval was obtained from SingHealth Centralized Institutional Review Board (2015/2011).

| Study population
Recruitment was carried out at outpatient oncology clinics. Patients who were newly referred to a medical oncologist and fulfilled the eligibility criteria were referred to the study team by their physicians.
The inclusion criteria were: (a) between 15 and 39 years old and (b) an official cancer diagnosis. Those who were unable to provide informed consent, had serious comorbid illnesses (such as neuropsychiatric conditions, which would severely affect quality of life), or would not have any scheduled visits subsequently were ineligible. For the purpose of this analysis, patients with organic causes of cognitive decline, such as primary and secondary CNS tumors, were excluded. All participants provided written informed consent prior to joining the study.

| Data collection
Participants were assessed at four time points over 1 year: (a) baseline at recruitment (T1); (b) 1 ± 0.5 month later (T2); (c) 6 ± 2 months later (T3); and (d) 12 ± 2 months later (T4  Scores of negatively worded items were reversed such that higher values indicated better self-perceived cognitive function. The total score was obtained by summing scores from all subscales. A decrease of more than 10.6 in total score has previously been established to indicate clinically important deterioration among Asian breast cancer patients. 16 The questionnaire developer has also recommended presenting data from individual subscales and using the PCI score as one of the primary measures, 17 where a score of less than 60 has been used to identify CRCI cases. 18 As suggested by the questionnaire developer, four items (MT1, MT2, PMT1, and PMT2) were only included in the score calculation after they were evaluated to fit with the scales, as indicated by Cronbach's alpha values of >.9 and item-to-scale correlation >.7.

| Distress and symptom burden
Distress thermometer (DT) is a brief screening tool that measures distress.
A problem checklist is also provided for respondents to identify items that have posed difficulties for them in the past week, including memory or concentration. The Rotterdam Symptom Checklist (RSCL) is used to assess the symptom burden of cancer patients. In this study, RSCL was also used to measure anxiety/depressive symptoms 19 and significant fatigue. 20 Details of both tools are provided in Supporting Information S1.

| Subjective CRCI
The mean change in FACT-Cog total and subscale scores were small in magnitude at all follow-up time points and mostly indicated improvement of cognitive function over time (

| Factors associated with self-reported cognitive function
After controlling for time, the mixed-effects model analysis showed that none of the treatment modalities (chemotherapy, surgery, or radiotherapy) were significantly associated with self-reported cognitive function. Based on model diagnostics (Supporting Information S5), gender, ethnicity, and smoking status were included in the final model. In our analysis, female gender (P = .02), Indian descent (P < .01), and smoking (P < .01) were associated with a higher T A B L E 3 Factors associated with cognitive function (measured based on total FACT-Cog score) among AYA cancer patients (N = 91) likelihood of reporting cognitive-associated complaints. Psychosocial factors, such as anxiety/depressive symptoms (P < .01) and fatigue (P < .01), both measured using the RSCL, were also associated with worse self-perceived cognitive function (Table 3). Age and education level were found to be nonsignificant predictors of self-perceived cognitive function and were excluded from the final model (Supporting Information S5). The statistical significance of associations remained unchanged when the PCI subscale score was used as the outcome measure to represent subjective cognitive function (Supporting Information S6).

| DISCUSSION
In this longitudinal study, one-third of the participants experienced self-reported CRCI at least once within a span of 12 months, despite their relatively young age and absence of CNS-associated cancer.
Compared to previous studies investigating subjective CRCI, the proportion of patients with subjective CRCI in our study was lower than that observed among breast cancer patients receiving chemotherapy (20%-50%) 21-23 but higher than patients who did not undergo chemotherapy (10%-15%). 21 In contrast to other studies, the mean deterioration in FACT-Cog total and subscale scores observed in our study were smaller in magnitude. Among breast cancer patients, mean changes of −10.4 and −6.5 in FACT-Cog total and PCI subscale scores, respectively, were reported at completion of chemotherapy. 21 Similarly, colorectal cancer patients reported a decrease of 9.4 and 5.7 in FACT-Cog total scores at 6 and 12 months after initiation of chemotherapy. 24 These observations suggest that CRCI is a less prominent problem in the AYA cancer population as a whole.
Nevertheless, in view of the significant proportion of AYA cancer patients who experienced clinically important CRCI, it is vital to identify individuals who are susceptible to CRCI. This is because cancerrelated complications have been suggested to reduce the performance of AYA cancer survivors in mental tasks at work, 10 translating into productivity loss of USD$2250 annually per AYA survivor in the United States. 12 The three demographic characteristics that have been shown to be associated with self-reported cognitive function in our study are consistent with reports in the literature. Similar to results reported from the LIVESTRONG survey 25 and an adult colorectal cancer cohort, 26 female patients were more likely than male patients to report cognitive problems. Ethnic differences in CRCI are also not unexpected given that genetic factors have been shown to be associated with varying degrees of susceptibility to self-reported CRCI. 22,23,27 Lastly, although smoking status has not been demonstrated to be linked to self-reported CRCI, a smoking history of 10 pack years or more has been reported to be associated with poorer intellectual function among newly diagnosed head and neck cancer patients measured using objective neurocognitive tests. 28 Interestingly, higher baseline FACT-Cog scores were noted among patients who exhibited CRCI at subsequent follow-up time points. This suggests that patients who eventually demonstrated significant CRCI were likely to have fewer cognitive complaints or better self-reported cognitive function at baseline. It has previously been found (albeit among breast cancer patients) that a low, rather than high, cognitive reserve increases the risk of developing CRCI. 29 Therefore, a more plausible explanation for our observation could be that AYA patients with high-performing roles have a lower tendency to report cognitive issues but are more sensitive to cognitive changes due to the nature of their work and daily routine. As a result, this group of patients initially had fewer cognitive complaints but was more likely to perceive CRCI over time. Alternatively, this trend could merely be an example of regression toward the mean. Nevertheless, further investigation is warranted by comparing these observed trends improve on limitations of this study. An age-matched noncancer control arm will be included to determine if the cognitive deterioration observed can be attributed to causes unrelated to cancer. Objective cognitive outcomes of the study participants will also be evaluated in the learning and memory, processing speed and executive function domains. This will provide further insight into specific brain areas that are affected by CRCI and illustrate whether the underlying pathology of CRCI is different in AYA cancer patients compared to other age groups.

| CONCLUSIONS
Our study addressed an important and neglected research gap within CRCI research. Although patients were relatively young and without CNS disease involvement, one-third of AYA cancer patients experienced significant decline in cognitive function. Demographic characteristics, such as gender, ethnicity, and smoking status were also found to be linked to poor self-reported cognitive function. Nevertheless, given the exploratory nature of our analysis, future work is needed with adequately powered studies which incorporate noncancer controls and neuropsychological assessments.