Second generation of soluble transferrin receptor assay – consequences for the interpretation of the ‘Thomas plot’

The ‘Thomas plot’ is of diagnostic value for distinguishing functional iron deficiency from classic iron deficiency in particular in patients with inflammatory conditions and cancer [1]. Therefore, the ‘Thomas plot’ became an extremely helpful and widely used diagnostic tool for clinicians to assess, monitor and treat the iron status of patients [2]. Four parameters need to be measured: C reactive Protein (CRP), serum ferritin, the hemoglobin content of reticulocytes (Ret-He) and the soluble transferrin receptor (sTfR). The calculated ferritin index (sTfR/lg Ferritin) is finally plotted against Ret-He [3]. Cut-offs for the ferritin index were initially defined dependent on the inflammatory status (evaluated by CRP) and the method used for sTfR-measurement. Especially the method used to determine sTfR is of great relevance as the assays of different vendors are not comparable in absolute values [3]. In 2021, Roche Diagnostics launched a new assay for sTfR measurement. Here we evaluated the comparability of the latest generation of the sTfR assay with the first generation and the consequences for cutoffs used in the Thomas plot.

To compare the two generations of sTfR assays, 69 patient samples were randomly selected and measured with both methods. The sTfR values of the second generation sTfR assay (range: 0.53–11.14 mg/L; median: 3.72 mg/L) were consistently lower compared to the first generation assay (range: 0.60–11.44 mg/L; median: 4.16 mg/L) (Figure 1A). Accordingly, also the ferritin indices (sTfR/lg Ferritin) were around 15 % lower when applying the values of the second generation sTfR assay (Figure 1B). However, both parameters show a good correlation.

Figure 1:

Passing-Bablok regression (upper panel) and Bland-Altman plots (lower panel) of the sTfR concentration (A) and the calculated ferritin index (B) using the first and second generation sTfR assay. Sixty nine patient samples were randomly selected and sTfR concentrations were determined using the first and second generation of Roche sTfR assay on a Cobas c701. The calculated ferritin indices were compared using the measured sTfR concentrations from panel A.

The fact that the second generation assay leads to lower concentrations for the sTfR is also reflected in the lower reference values of the second generation assay, provided by the manufacturer (first generation: 2.2–5.0 mg/L (male), 1.9–4.4 mg/L (female); second generation: 1.71–4.13 mg/L). Figure 2 illustrates the problems putatively associated when using the same cut-offs with the two assays. As the second generation assay lead to lowered sTfR concentration and in consequence to lower ferritin indices, patients with marginal ferritin indices would be falsely diagnosed using the ‘Thomas’ plot when applying the old cut-offs. This would lead to a switch from quadrant 1 into 2 in case of Ret-He above 1.74 fmol. For Ret-He below 1.74 fmol this would possibly lead to switch from quadrant 4 to 3 (Figure 2). In both cases this might lead to false therapeutic decisions (e.g. not necessary Erythropoitin administration, not treating or discontinue treatment with iron replacement products).

Figure 2:

Consequences of lowered sTfR concentrations measured with the second generation sTfR assay from Roche when using the originally cut-offs in dependence of patient’s CRP value. Due to the lowered sTfR concentrations using the second generation assay the iron reserves would be overestimated and iron deficiency could be overlooked. The grey area represents the range of Ferritin-Index, which is critical due to the new sTfR-assay (CRP<5 mg/L: 3.2<[Ferritin-Index]<3.7 mg/L and for CRP>5 mg/L: 2.0<[Ferritin-Index]<2.3 mg/L). In our retrospective analysis of patients in our maximum care hospital 5.2 % of all Ferritin-indices would be in the critical range (4.4 % with switch from Q1 to Q2 and 0.8 % with switch from Q4 to Q3). For details see Table 1.

To get an idea how many patients would be affected, we retrospectively analyzed the cases between the year 2016 and 2021 in our maximum care hospital (university hospital). In total, 15,592 patients were included independently from the anemic status. From the results of the linear correlation we extrapolated the consequences when using the new assay with the old cut-offs and identified that 813 (5.2 %) results of the Thomas plot would be affected. Table 1 shows that most of the affected results would lead to a switch from Q1 to Q2, but this is also in accordance to the included patients from our hospital. In addition, we calculate reference intervals for sTfR from our intrahospital patients using the refineR method [4]. The reference interval for sTfR first generation calculated from our data was 1.60–6.36 mg/L (calculated from 12,718 data points) and 1.49–6.11 mg/L (calculated from 8,390 datapoints) for the sTfR second generation assay (distribution curves can be found in Supplementary Material). In a recent study of Lyle et al. the differences in the sTfR values between the first and second generation of the Roche sTfR assay were also seen [5]. In addition, this paper highlights another problem of the sTfR assays from different manufacturers which is the lack of standardization. Although a WHO standard is available, only the minority of assays are calibrated against this standard, which might be an ideal source of error for misinterpretation of sTfR values.

In summary, due to the differences in sTfR-levels in the first and second generation assays, the Ferritin-Index can be falsely interpreted when using the old cut-offs. We suggest to adjust the Ferritin-Index cut-offs. According to our results we suggest reducing the originally published cut-off levels from Thomas et al. from 2 to 1.8 or from 3.2 to 2.8, respectively [3]. Afterwards, we recommend that each laboratory should verify whether the original cut-off levels are still be applicable in addition to the check of the own patient cohort [6].