Low-cost transcriptional diagnostic to accurately categorize lymphomas in low- and middle-income countries

Abstract

Inadequate diagnostics compromise cancer care across lower- and middle-income countries (LMICs). We hypothesized that an inexpensive gene expression assay using paraffin-embedded biopsy specimens from LMICs could distinguish lymphoma subtypes without pathologist input. We reviewed all biopsy specimens obtained at the Instituto de Cancerología y Hospital Dr. Bernardo Del Valle in Guatemala City between 2006 and 2018 for suspicion of lymphoma. Diagnoses were established based on the World Health Organization classification and then binned into 9 categories: nonmalignant, aggressive B-cell, diffuse large B-cell, follicular, Hodgkin, mantle cell, marginal zone, natural killer/T-cell, or mature T-cell lymphoma. We established a chemical ligation probe-based assay (CLPA) that quantifies expression of 37 genes by capillary electrophoresis with reagent/consumable cost of approximately $10/sample. To assign bins based on gene expression, 13 models were evaluated as candidate base learners, and class probabilities from each model were then used as predictors in an extreme gradient boosting super learner. Cases with call probabilities < 60% were classified as indeterminate. Four (2%) of 194 biopsy specimens in storage <3 years experienced assay failure. Diagnostic samples were divided into 70% (n = 397) training and 30% (n = 163) validation cohorts. Overall accuracy for the validation cohort was 86% (95% confidence interval [CI]: 80%-91%). After excluding 28 (17%) indeterminate calls, accuracy increased to 94% (95% CI: 89%-97%). Concordance was 97% for a set of high-probability calls (n = 37) assayed by CLPA in both the United States and Guatemala. Accuracy for a cohort of relapsed/refractory biopsy specimens (n = 39) was 79% and 88%, respectively, after excluding indeterminate cases. Machine-learning analysis of gene expression accurately classifies paraffin-embedded lymphoma biopsy specimens and could transform diagnosis in LMICs.

The Impact of Artificial Intelligence on Health Equity in Oncology: Scoping Review

“…In addition, our review identified equity issues related to the cost of AI technologies; such technologies can be costly and may not be affordable in many LMICs under existing economic circumstances. Love et al [ 102 ] developed an AI device to triage breast lumps in low-resource settings but noted that “the device used in this study is more expensive than most LMICs settings can afford, lower cost devices are becoming more available.” However, others were able to create technologies that may be more affordable for LMICs: a gene expression assay costing US $450 capable of assessing samples for only US $10 [ 130 ].…”

“…Moreover, it is important to recognize that the inclusion of researchers from LMICs in the design of AI technologies alone does not guarantee widespread improvements in health for patients in these countries. Rather, benefits are often limited to select partner sites of HICs; therefore, while these technologies may help address global disparities, they may exacerbate inequities within LMICs [ 130 ]. To ensure a more equitable distribution of benefits within LMICs, research should extend beyond specific partner institutions, engaging additional stakeholders from relevant government and nongovernmental organizations to evaluate and implement technologies.…”

“…One additional problem that has been raised surrounding the use of AI in LMICs is the issue of data colonialism [ 160 ], a practice in which data are extracted from LMICs by institutions in HICs for the purposes of building algorithms whose benefits accrue primarily to stakeholders in HICs [ 161 ]. Although articles from our review did not engage directly with these issues, some did discuss important considerations for what collaboration means between HICs and LMICs [ 85 , 97 , 130 , 131 ]. However, there was limited acknowledgment of ethical issues arising from the involvement of LMICs as mere resources for data extraction and algorithmic training or as an exploratory ground for novel applications of AI technologies for global health.…”

“…Without prioritizing investment in the basic infrastructure, such as appropriate hardware to run AI programs, buildings where such hardware can be housed, and cables to carry digital signals, the utility of these technologies in the global health context should be questioned [ 172 , 173 ]. A number of articles identified in our review engaged with these voiced concerns, with some researchers creating technologies with the infrastructural capacities of specific LMICs in mind and others highlighting the need for additional infrastructure to support the technology they developed [ 87 , 102 , 129 , 130 , 148 ].…”

“…Love et al [102] developed an AI device to triage breast lumps in low-resource settings but noted that "the device used in this study is more expensive than most LMICs settings can afford, lower cost devices are becoming more available." However, others were able to create technologies that may be more affordable for LMICs: a gene expression assay costing US $450 capable of assessing samples for only US $10 [130].…”