Symptom trajectories in patients with cancer: a scoping review of person-centered trajectory modeling studies
Year: 2026
Type: preprint
Abstract: Patients with cancer experience multiple concurrent symptoms that cluster, co-occur, and fluctuate across the illness trajectory — from diagnosis through treatment to survivorship or advanced disease [L1]. Symptom cluster research has established that these symptoms are not independent, and that their combined burden is a stronger predictor of functional status, quality of life, and treatment adherence than any single symptom [L2]. Most longitudinal symptom research, however, summarizes change at the group level, using variable-centered approaches (e.g., repeated-measures ANOVA, mixed-effects models, generalized estimating equations) that estimate a single average trajectory and assume that patients respond to illness and treatment in a broadly homogeneous way [L3]. This assumption obscures clinically meaningful heterogeneity: subgroups of patients may follow distinct symptom trajectories (e.g., persistently high, rapidly escalating, recovering, or resilient patterns), and identifying these subgroups is a prerequisite for targeted, stratified supportive care. Person-centered (person-oriented) trajectory modeling addresses this gap. This family of latent variable mixture methods — including latent class growth analysis (LCGA), growth mixture modeling (GMM), group-based trajectory modeling (GBTM), and latent transition analysis (LTA) — identifies unobserved subpopulations characterized by distinct longitudinal symptom patterns [L4]. Applications of these methods in cancer symptom research have grown rapidly over the past decade. This growth, however, has outpaced methodological standardization. Published studies vary widely in how they enumerate and select the number of latent classes (e.g., fit indices such as BIC, aBIC, entropy, and the bootstrap likelihood ratio test, versus interpretability and parsimony), how they handle missing longitudinal data, how they treat covariates and distal outcomes (e.g., one-step versus three-step approaches), how they validate and name the identified subgroups, and whether the resulting trajectory classes are ever translated into clinical follow-up decisions [L5]. Without a systematic map of these practices, the strength of conclusions drawn from this literature — and the feasibility of the stratified care it promises — remains difficult to assess. To our knowledge, no review has synthesized this literature for symptom outcomes in cancer. Existing reviews cover adjacent but distinct territory: a 2025 systematic review addressed network analysis of symptoms (a cross-sectional, person-centered method, but not trajectory modeling) [L6a]; and 2025 systematic reviews of longitudinal latent class modeling were limited to specific non-symptom outcomes — supportive care needs (PMID: 40678771) and fear of cancer recurrence (PMID: 38584241) [L6b]. No scoping review has mapped the full body of person-centered trajectory modeling studies across symptom outcomes (symptom clusters, symptom burden, and individual core symptoms) in patients with cancer. Aim and research questions This scoping review aims to map how person-centered trajectory modeling has been applied to symptom outcomes in patients with cancer, and to identify methodological and clinical gaps to inform future longitudinal symptom research. The review addresses the following research questions: 1. RQ1. In which cancer populations, treatment phases, follow-up windows, and symptom outcomes have person-centered trajectory modeling methods (LCGA, GMM, GBTM, LTA, and closely related mixture models) been applied? 2. RQ2. How have studies determined the number of latent classes and the adequacy of the final model (statistical criteria, interpretability criteria, sample size and class-size considerations)? 3. RQ3. How have studies handled missing longitudinal data, attrition, and irregular measurement occasions, and how have covariates and distal outcomes been incorporated (one-step, three-step, or other approaches)? 4. RQ4. What trajectory subgroups have been identified (number, shape, and naming of classes), and to what extent have the identified subgroups been validated or linked to clinical outcomes and care decisions? 5. RQ5. What methodological and clinical gaps exist, and what priorities emerge for future person-centered symptom trajectory research in oncology? Consistent with the purpose of a scoping review, no assumptions are made about the direction or quality of the evidence; the review will report the state of the field as it stands, including heterogeneity and reporting deficiencies where they exist.
Source: OSF Preprints (OSF Preprints)
Author: YuXiao Zhu
Language: English
Topic: Cancer survivorship and care
Subfield: Oncology
Field: Medicine
Domain: Health Sciences
Open Access status: green
OpenAlex ID: W7224768139