Objectives Trichloroethylene (TCE) and Perchloroethylene (PER) are two chlorinated solvents that have been applied widely as degreaser of metal parts and in dry cleaning and other applications. In 2012 the International Agency for Research on Cancer classified TCE as carcinogenic to humans and PER as probably carcinogenic to humans. We explored exposure-response relations for TCE and PER and non-Hodgkin lymphoma (NHL), multiple myeloma (MM), and cancers of the kidney and liver in the Nordic Occupational Cancer (NOCCA) cohort which holds information on occupation and long-term follow-up for cancer for the populations of Norway, Sweden, Finland and Iceland.

Methods The cohort was set-up by linking occupational information from censuses to national cancer registry data utilising personal identity codes in use in all the Nordic countries. Country, time-period, and job-specific exposure estimates were generated for TCE, PER, and potentially confounding occupational exposures with a comprehensive job-exposure matrix. Two exposure metrics were created: cumulative exposure and ‘average intensity times prevalence’ of exposure. Conditional logistic regression was conducted for exposure groups as well as for continuous cumulative exposure.

Results Hazard ratios for liver cancer, NHL, MM, but not kidney cancer were slightly elevated in groups with high exposure to PER (compared to occupationally unexposed subjects). Hazard ratios for liver cancer and NHL also increased with increasing continuous exposure to PER. We did not observe evidence for an association between exposure to TCE and NHL, MM, and liver and kidney cancer.

Conclusions Although this study was subject to limitations related to the low prevalence of exposure to PER and TCE in the Nordic population and a necessarily rudimentary exposure assessment strategy, we observed some evidence indicative of an excess risk of cancer of the liver and NHL in subjects exposed to PER. Exposure misclassification may have diluted the observed associations in this study.