Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2024 Oct;13(19):e70278.
doi: 10.1002/cam4.70278.

Symptom Network Analysis and Unsupervised Clustering of Oncology Patients Identifies Drivers of Symptom Burden and Patient Subgroups With Distinct Symptom Patterns

Brandon H Bergsneider  1   2 , Terri S Armstrong  1 , Yvette P Conley  3 , Bruce Cooper  4 , Marilyn Hammer  5 , Jon D Levine  6 , Steven Paul  4 , Christine Miaskowski  4   6 , Orieta Celiku  1
Affiliations

Symptom Network Analysis and Unsupervised Clustering of Oncology Patients Identifies Drivers of Symptom Burden and Patient Subgroups With Distinct Symptom Patterns

Brandon H Bergsneider et al. Cancer Med. 2024 Oct.

Abstract

Background: Interindividual variability in oncology patients' symptom experiences poses significant challenges in prioritizing symptoms for targeted intervention(s). In this study, computational approaches were used to unbiasedly characterize the heterogeneity of the symptom experience of oncology patients to elucidate symptom patterns and drivers of symptom burden.

Methods: Severity ratings for 32 symptoms on the Memorial Symptom Assessment Scale from 3088 oncology patients were analyzed. Gaussian Graphical Model symptom networks were constructed for the entire cohort and patient subgroups identified through unsupervised clustering of symptom co-severity patterns. Network characteristics were analyzed and compared using permutation-based statistical tests. Differences in demographic and clinical characteristics between subgroups were assessed using multinomial logistic regression.

Results: Network analysis of the entire cohort revealed three symptom clusters: constitutional, gastrointestinal-epithelial, and psychological. Lack of energy was identified as central to the network which suggests that it plays a pivotal role in patients' overall symptom experience. Unsupervised clustering of patients based on shared symptom co-severity patterns identified six patient subgroups with distinct symptom patterns and demographic and clinical characteristics. The centrality of individual symptoms across the subgroup networks differed which suggests that different symptoms need to be prioritized for treatment within each subgroup. Age, treatment status, and performance status were the strongest determinants of subgroup membership.

Conclusions: Computational approaches that combine unbiased stratification of patients and in-depth modeling of symptom relationships can capture the heterogeneity in patients' symptom experiences. When validated, the core symptoms for each of the subgroups and the associated clinical determinants may inform precision-based symptom management.

Keywords: Gaussian graphical models; network analysis; precision medicine; quality of life.

PubMed Disclaimer

Conflict of interest statement

The authors declare no conflicts of interest.

Figures

FIGURE 1
FIGURE 1
Severity of the 32 MSAS symptoms across all 3088 cancer patients and survivors: (A) Symptom severity distributions. (B) Percentage of patients reporting mild‐very severe symptoms (severity ≥ 1) and (C) severe‐very severe symptoms (severity ≥ 3).
FIGURE 2
FIGURE 2
In the network for all patients and survivors, symptoms cluster into three groups: (A) A regularized Gaussian Graphical Model (GGM) was generated from the symptom severity data from all of the patients. Nodes represent symptoms. Edges represent the partial correlation between the symptom severities. Edge thickness is proportional to the strength of the correlation. Lack of edges between symptoms indicates that the regularization procedure deemed their correlations too weak to represent. All correlations were positive. Three symptom clusters were identified: Constitutional (red), gastrointestinal‐epithelial (green), and psychological (blue). (B) Symptom stability was assessed by reconstructing the network for random permutations of the dataset and calculating the proportion of times each symptom occurred in the symptom cluster shown in part A. (C) Z‐scored strength, closeness, and betweenness centrality for all symptoms in the network, ordered from highest to lowest according to strength centrality. (D) Z‐scored bridge strength, closeness, and betweenness for each symptom in the network, ordered from highest to lowest according to bridge strength centrality.
FIGURE 3
FIGURE 3
Second‐order unsupervised clustering identifies six patient subgroups with unique symptom severity patterns: (A) Heatmap of symptom severities for each patient subgroup. Symptoms are ordered based on the symptom clusters identified in Figure 2. (B–G) Average severity of each symptom in each patient subgroup (blue = average symptom severity for subgroup, red = average symptom severity for all patients). 95% confidence intervals are shown for each symptom.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

References

    1. Harris C. S., Kober K., Cooper B., et al., “Symptom Clusters in Oncology Outpatients: Stability and Consistency Across a Cycle of Chemotherapy,” BMJ Supportive & Palliative Care 13 (2023): e1198–e1211, 10.1136/spcare-2022-003785. - DOI - PMC - PubMed
    1. Miaskowski C., Paul S. M., Harris C. S., et al., “Determination of Cutpoints for Symptom Burden in Oncology Patients Receiving Chemotherapy,” Journal of Pain and Symptom Management 63, no. 1 (2022): 42–51, 10.1016/j.jpainsymman.2021.07.018. - DOI - PMC - PubMed
    1. Zhu Z., Sun Y., Kuang Y., et al., “Contemporaneous Symptom Networks of Multidimensional Symptom Experiences in Cancer Survivors: A Network Analysis,” Cancer Medicine 1 (2022): 663–673, 10.1002/cam4.4904. - DOI - PMC - PubMed
    1. Shin J., Hammer M., Cooley M. E., et al., “Common and Distinct Risk Factors That Influence More Severe and Distressing Shortness of Breath Profiles in Oncology Outpatients,” Cancer Medicine 13, no. 3 (2024): e7013, 10.1002/cam4.7013. - DOI - PMC - PubMed
    1. Sorrera D., Block A., Mackin L., et al., “Decrements in Both Physical and Cognitive Function Are Associated With a Higher Symptom Burden in Oncology Patients,” Seminars in Oncology Nursing 39, no. 6 (2023): 151516, 10.1016/j.soncn.2023.151516. - DOI - PubMed