In men who tested positive for HLA-DRB1*15:01, there was a suggestive association between exposure to pesticides and multiple sclerosis (MS), according to research published in Multiple Sclerosis and Related Disorders.
Encouraged by recent studies that used machine learning technologies to increase accuracy in predictive models, the study authors opted to use the least absolute shrinkage and selection operator regression (LASSO) technique to stratify potentially relevant environmental exposures in a population-based case-control study using data from Kaiser Permanente’s Northern California member base. Investigators also replicated details of the study using data from the Swedish Epidemiological Investigation in Multiple Sclerosis. Researchers reviewed the medical records of 2310 people with MS and 4819 matched controls.
Participants completed questionnaires that collected information about their lifestyle and various exposures. Environmental exposures evaluated in the research include those related to animals, diet, recreation, and occupation; examples include consumption of meats, pet exposure, nail polish exposure, pesticide/herbicide exposure, caffeine consumption, and cigarette smoking.
Results of the study revealed no meaningful associations related to MS; however, researchers identified a strong trend toward an association for men exposed to pesticides with higher odds observed in those who were positive for HLA-DRB1*1501 (odds ratio pooled 3.11; 95% CI, 0.87-11.16; P =.08).
The study limitations included the self-reporting of exposures and a retrospective design that introduced recall bias. The study also suffered from poor generalizability due to smaller numbers of men in the participant pool and low numbers of non-white populations.
Researchers propose that future studies include much larger participant pools and prospective designs. Investigators also note that although the study findings require confirmation, the study demonstrated “the application of LASSO to identify environmental exposures with reduced multiple statistical testing penalty” and “[machine] learning approaches may be useful for future investigations of concomitant MS risk or prognostic factors.”
Mowry EM, Hedström AK, Gianfrancesco MA, et al. Incorporating machine learning approaches to assess putative environmental risk factors for multiple sclerosis. Mult Scler Relat Disord. 2018; 24:135-141.