We explored several risk factors for predicting CIPN symptom severity after 6 weeks of neurotoxic taxane or platinum chemotherapy in women with breast cancer by assessing pre-chemotherapy inflammatory, behavioral, clinical and psychosocial risk factors. We found several consistent risk factors for CIPN including more severe baseline neuropathy, more severe fatigue/anxiety/depression, lower levels of anti-inflammatory IL-10, and black race.
This is the largest and among the first study in human patients with cancer (not preclinical models, e.g. [33]) suggesting that a pro-inflammatory state before chemotherapy is a risk factor for CIPN. We found that lower levels of anti-inflammatory IL-10 and higher levels of pro-inflammatory IFN-γ in serum were both risk factors for more severe CIPN, based on both patient-reports for the primary outcome numbness and tingling and the secondary outcome hot/coldness in hands/feet. Other risk factors included higher pro-inflammatory IL-1β (for numbness and tingling), pro-inflammatory IL-8 (for numbness/tingling), and lower IL-6 (for hot/coldness), which has both pro- and anti-inflammatory properties [51]. Our results are consistent with several prior studies: (1) one suggesting that blocking IL-1β signaling reduced CIPN symptoms in rats who received paclitaxel [52]; (2) a correlational study in humans suggesting that a single IL-1β polymorphism is a risk factor for CIPN [53]; and (3) a case-control study in humans suggesting perturbed gene expression in neuroinflammatory pathways in breast cancer survivors with CIPN vs. those without CIPN [54]. Our observations lend support and add specificity to theories that inflammation is involved in the etiology of CIPN in humans [26-28, 55], and can help advance and inform anti-inflammatory treatments for CIPN [56, 57].
Our results suggest that one of the strongest risk factors for CIPN symptom severity is the symptom cluster of fatigue, anxiety, and depression (typically accounting for 18% variance), consistent with prior research [13, 15]. These results invite the hypothesis that there is a shared biological mechanism of fatigue, anxiety, depression, and CIPN; the mechanism might be peripheral inflammation, neuroinflammation via disrupted blood-brain-barrier integrity [58], changes in the brain [59-61] including the interoceptive brain system, which processes bodily sensations [62], or mitochondrial dysfunction [63], all of which are implicated in the etiology of not only CIPN (for discussion, see [38]) but also fatigue, anxiety, and depression [63-65]. Furthermore, if fatigue, anxiety, and depression play causal roles in the development of CIPN symptoms, then interventions to reduce fatigue, anxiety, and depression could help prevent or alleviate symptoms of CIPN [66]. Potential interventions include (1) exercise, which improves depression [67], anxiety [67], fatigue [68], inflammation [69], and affects the interoceptive brain system (e.g., [70]), and may reduce CIPN [38, 39] (for a review, see [66, 71-73]); (2) pharmaceuticals such as duloxetine, which improves depression [74] and CIPN [75]; (3) neuromodulation or neurofeedback, which improves depression [76-78] and may also improve CIPN (e.g., [79]).
Our weakest observed risk factors for CIPN include age and the use of diabetes medications, consistent with mixed findings in the literature (see Introduction for citations). The mixed findings may be due to statistically weak associations and across-study differences in samples, sample sizes, use of a cutoff vs. continuous measures for age, and choice of other covariates. Some of the risk factors for CIPN that we observed might be due to confounding chemotherapy factors such as type and dose, which we did not collect. Physical activity (pedometer steps) and BMI were both weak or non-significant predictors of CIPN severity despite prior studies indicating that physical activity is protective [16-18] and BMI is a risk factor [16, 20]. Our unexpected observations may be because our eligibility criteria ensured that our sample was sedentary (low steps, high BMI), thus restricting variability in physical activity levels. Our observation that black patients were at greater risk for CIPN severity was consistent with prior work [16], and is perhaps due to undiagnosed or untreated comorbidities such as prediabetes or diabetes, which is more prevalent in black Americans compared to white Americans [80-82].
This study has several strengths. First, to our knowledge, this is the largest longitudinal human study of inflammation and CIPN (116 patients, with 55 patients providing blood for inflammation measures; the only prior study involved 17 patients [34]). Second, assessing CIPN before and after six weeks of chemotherapy provides stronger evidence for CIPN risk factors than a cross-sectional post-chemotherapy design. Third, our data were obtained from community oncology clinics around the United States, enhancing the generalizability to our findings compared to single-site studies.
This study also has limitations. First, because this was an exploratory secondary analysis [37] based on subpopulation of 116 patients randomized to usual care from a randomized clinical trial, these findings should be tested for replication and further validation. With that said, our pattern of findings with pro- and anti-inflammatory cytokines are as expected based on prior literature and presumed mechanisms, lending credibility to these results. Second, the measures of CIPN were crude, with two 0-10 patient-reported questions instead of a full multi-question survey designed to assess CIPN. However, the measures we used have been successfully used and validated in prior studies (e.g., [15, 43]). Moreover, we did not have access to clinical assessment of CIPN (e.g., monofilament test, nerve conduction) specific to CIPN. Third, we only measured the first 6 weeks of chemotherapy and therefore the severity of neuropathy in our sample was relatively low. However, our 6-week measure may serve as a surrogate for future CIPN severity, as paclitaxel causes a monotonic weekly increase in CIPN severity (e.g., Figure 4 from [83]). Fifth, our precision may be reduced given small cell sizes in some variables (e.g., only a few black patients). In addition, we do not have information on certain variables that might be related to neuropathy symptoms, such as smoking status, alcohol consumption, HIV status, and chemotherapy dose. Sixth, our assessment of serum cytokines may not necessarily reflect inflammation in the nervous system that contributes to CIPN. However, serum inflammation is convenient to assess and could serve as a useful biomarker of CIPN in the future. In addition, detection limits of cytokine assessment methodology forced us to use categorical coding for some cytokines (e.g., IL-1β, IFN-γ). However, we still saw prediction of CIPN severity with cytokines that were not subject to categorical coding (e.g., IL-10, IL-6). Finally, it is not known how these results generalize to other populations, as our sample was primarily white, well-educated, female patients with breast cancer who were sedentary and reported mild (not severe) neuropathy.