UCSF

Oncology patients receiving chemotherapy report on average 14 concurrent symptoms. The co-occurrence of these symptoms is associated with poorer functional status, decrements in quality of life (QOL), and increased mortality. Given that symptoms rarely occur in isolation, the concept of a symptom cluster emerged in the literature in 2001. An increased understanding of how symptoms cluster together and the biological mechanism(s) that underlie them has the potential to lead to the development of targeted interventions to decrease symptom burden. Therefore, the overall aims of this dissertation research were to: 1) review the conceptual basis for using variable-centered versus patient-centered analytic approaches in symptom cluster research; 2) systematically review studies published since 2016 that evaluated for symptom clusters in patients receiving primary or adjuvant chemotherapy; 3) evaluate the stability and consistency of symptom clusters across time and across three symptom dimensions (i.e., occurrence, severity, and distress); 4) identify common and distinct symptom clusters across various types of cancer; and 5) evaluate for associations between psychological and gastrointestinal symptom clusters and epigenetic regulation of inflammatory genes in a heterogeneous sample of oncology patients. In terms of aim 1, a theoretical paper described two conceptual approaches that are used to evaluate symptom clusters; namely: “clustering” symptoms (i.e., variable-centered analytic approach) and “clustering” patients (i.e., person-centered analytic approach). Findings suggest that while each approach has unique strengths and weaknesses, conceptual clarity is needed when a study is designed and the specific research question(s) should guide the selection of the appropriate analytic method. The application of newer analytic approaches (e.g., network analysis (NA), natural language processing (NLP)) to study symptom clusters were reviewed. This paper summarized the paucity of research on the evaluation of the underlying mechanism(s) for symptom clusters. In terms of aim 2, in a systematic review, 23 studies were identified that evaluated for symptom clusters in patients receiving chemotherapy. Across these studies, the Memorial Symptom Assessment Scale (MSAS) was the most common instrument and exploratory factor analysis (EFA) was the most common statistical method used to identify symptom clusters. While psychological, gastrointestinal, and nutritional clusters were the most common clusters identified across studies, only the psychological cluster remained relatively stable over time. A major conclusion from this review was that clear criteria are needed to evaluate the stability of symptom clusters across time and dimensions. In addition, only five studies evaluated for secondary outcomes (e.g., functional status, QOL). Additional research is needed to evaluate the biological mechanism(s) for symptom clusters. In terms of aim 3, prior to the start of their second or third cycle of chemotherapy, outpatients reported an average of 13.9 (±7.2) concurrent symptoms. Lack of energy was both the most common and severe symptom while “I don’t look like myself” was the most distressing. Psychological, gastrointestinal, weight gain, respiratory, and hormonal clusters were the common symptom clusters identified across the three symptom dimensions. Our findings suggest that psychological, gastrointestinal, and weight gain clusters are common across various types of cancer while respiratory and hormonal clusters are cancer specific. In terms of aim 4, across a cycle of chemotherapy, the number of symptoms remained relatively stable over time, with patients reporting 13.9 (±7.2) symptoms prior to, 14.0 (±7.0) at one week after, and 12.2 (±6.8) at two weeks after receipt of chemotherapy. While the psychological, weight gain, respiratory, and gastrointestinal clusters were stable over time and dimensions, only the psychological, weight gain, and respiratory clusters were consistent across time and dimensions. In terms of aim 5, given the paucity of studies on the underlying mechanism(s) for the two most common symptom clusters (i.e., psychological, gastrointestinal), exploratory analyses were done to evaluate for associations between these clusters and epigenetic variation of inflammatory genes. Findings from both studies provide preliminary support for the hypothesis that epigenetic dysregulation of inflammatory processes contributes to the occurrence of psychological and gastrointestinal symptom clusters in patients receiving chemotherapy. For the psychological symptom cluster, cluster of differentiation (CD) 40 was differentially methylated across two independent samples (false discovery rate (FDR) = .017). Six expression-associated CpGs (i.e., eCpG; cg22232207, cg06571407, cg17929951, cg21601405, cg01943874, cg11841529) located in the promoter region of this gene were hypomethylated across both samples. For the gastrointestinal symptom cluster, one trans eCpG locus (i.e., cg03171795) that was associated with expression of the lymphotoxin beta (LTB) gene was associated with the occurrence of the gastrointestinal symptom cluster (FDR = 0.168). These findings warrant validation. This dissertation concludes with implications for clinical practice and future research.