A clinical predictive score of high liver iron content in metabolic hyperferritinemia: a retrospective cohort pilot study.

BACKGROUND

In metabolic hyperferritinemia, most patients do not require bloodletting as the liver iron content is mildly increased. We aimed to develop a clinical predictive score of high liver iron content in metabolic hyperferritinemia to guide the prescription of magnetic resonance imaging of the liver.

METHODS

We conducted a single-center retrospective cohort study including consecutive patients with metabolic hyperferritinemia who underwent a liver iron content evaluation at diagnosis. Excessive alcohol consumption was an exclusion criterion. A multivariate analysis followed by a 1000 bootstrap replicate analysis with an expectation-maximization algorithm was used to identify the predictive factors of high liver iron content. A ROC curve analysis was built to study the performance of the score based on the odds-ratio provided by the multivariate analysis.

RESULTS

217 patients (180 men, mean age 57 years old) were included. Fifty-five patients (25%) had high liver iron content (≥ 100 µmol/g). In univariate analysis, a family history of hyperferritinemia requiring phlebotomies was associated with high LIC, as well as an increase of transferrin saturation > 45% (p < 0.001). In multivariate regression, a family history of hyperferritinemia (OR 6.15, CI95 [2.11-17.92]), increased ferritin level ≥ 600 µg/L (OR 5.53, CI95 [1.43-21.42]) and increased transferrin saturation ≥ 45% (OR 2.63, CI95 [1.32-5.23]) were significantly associated with high liver iron content. A 15-point predictive score (area-under-the-curve 0.72, CI95 [0.64-0.79], p < 0.001) was built, providing an OR of 4.17 (CI95 [2.15-8.07], p < 0.001) for high liver iron content (sensitivity 60%, specificity 97%, negative predictive value 84%).

CONCLUSION

in this pilot study, ferritin ≥ 600 µg/L, transferrin saturation ≥ 45% and a family history of hyperferritinemia requiring bloodletting provided a simple clinical score to predict high liver iron content in metabolic adult hyperferritinemia. The bootstrap analysis confirmed the robustness of our model.