The Decisive Role of Pronounced Hyperferritinemia for Diagnosis and Prognosis in the Spectrum of Hematological Disorders

Background

Ferritin is an iron storage protein that can also be elevated in serum by various inflammatory pathologic states. Hyperferritinemia is defined as ferritin levels ≥500 ng/mL and is mainly related to hemophagocytic lymphohistiocytosis (HLH) when seen in association with hematologic abnormalities. However, its prevalence in the general population has been reported as high as 13%, and its diagnostic and prognostic role has not been determined among the different hematologic disease subgroups, nor for different levels besides extreme hyperferritinemia (ferritin levels ≥10,000 ng/mL). Our aim was to describe the hematologic etiologies for hyperferritinemia and define the impact over diagnosis and clinical outcomes of different ferritin values within the 500 ng/mL and 15,000 ng/mL range.

Methods

This was a retrospective single-center study that included all consecutive cases of hyperferritinemia according to our laboratory records between January 2015 and December 2019. A total of 6602 cases were identified. After excluding cases with incomplete medical records or redundant ferritin measurements, more than one measurement within six months, 3193 cases were included.

Chi-squared test and Kruskal-Wallis ANOVA tests were performed to demonstrate differences between categorical and quantitative variables, respectively. ROC curves were used to analyze the diagnostic performance of our tests. Survival analyses were carried out with Kaplan-Meier curves. All statistical analyses were carried out by IBM SPSS version 26.

Results

After non-HIV infection (26.1%; n=832), hematologic disease was the second most common etiology at 23.5% (n=749) within the cohort. The median age for this group was 50 years (17-95), and the median ferritin level was 1,133.3 ng/mL (500-15,000). Hematologic diagnoses were grouped into four categories: hematologic malignancy (60%), iron overload (26%), hemolytic anemia (7%), and HLH (7%). Median serum ferritin was higher in patients with HLH (6,717.5 ng/mL) when compared to other subgroups (p<0.01). Additional baseline characteristics are shown in Table 1.

With a cut-off of >3,000 ng/ml (pronounced hyperferritinemia), ferritin level as a diagnostic test for HLH had a sensitivity of 75%, specificity of 87.8%, an area under the curve (AUC) of 0.814, positive predictive value (PPV) of 31.5%, and negative predictive value (NPV) of 97.9%. The same cutoff value in all 3,193 episodes of hyperferritinemia had similar results (sensitivity of 75%, specificity of 92.5%, AUC of 0.837, PPV of 14.1%, NPV of 99.6%). There was no association between pronounced hyperferritinemia and hematologic malignancy, iron overload, or hemolytic anemia.

Overall survival (OS) at six months was 79.6% (n=596). When stratified per hematologic diagnosis, OS was 55.8% (n=29) for HLH, 76.4% (n=344) for hematologic malignancy, 89.8% (n=176) for iron overload, and 92.2% (n=47) for hemolytic anemia (p<0.01). On univariate analysis, pronounced hyperferritinemia was related to an increased mortality in patients with hematologic malignancy (p<0.01) and iron overload (p<0.01). There was no association between pronounced hyperferritinemia and mortality for HLH or hemolytic anemia (Figure 1).

Conclusion

To our knowledge, this is the first study to analyze the impact of hyperferritinemia in patients with diverse hematologic diagnoses. Our findings suggest that a cut-off value of >3000 ng/mL has a high NPV to exclude the diagnosis of HLH, within other differential diagnoses. Regarding prognosis, besides HLH, where clinical outcomes are known to be poor, pronounced hyperferritinemia can be useful in identifying patients with a higher risk of early mortality within the hematologic malignancy and iron overload subgroups.