To the Editor:

We read with great interest the article by Henke et al¹ describing that erythropoietin (Epo) treatment might adversely affect prognosis of head and neck cancer patients if cancer cells express erythropoietin receptors (EpoR).

In this article, the authors concluded that the locoregional progression-free survival was substantially poorer if epoetin beta was administered to patients showing EpoR expression compared with placebo.¹ Although the difference in treatment-associated relative risks was borderline statistically significant, the data are provocative and suggest that it is necessary to pay great attention in the use of recombinant Epo in the treatment of anemic cancer patients.¹

However, we have strong reservations about the specificity of the antibody to EpoR (C-20 anti-EpoR, sc-695, Santa Cruz Biotechnology, Santa Cruz, CA raised against the C-terminal cytoplasmic domain), that was used to evaluate EpoR presence in the investigation by Henke et al. Recently, Elliott et al² reported that this antiserum cannot be employed for investigating EpoR expression because it lacks of affinity and specificity. These antibodies seem to detect, almost exclusively, a band corresponding to the heat shock protein 70 (HSP70).² Thus, the article by Elliott et al casts serious doubts on the detection of EpoR on cancer cells by sc-695.

We have reinvestigated the EpoR topic and have obtained further data which demonstrated the lack of specificity of sc-695 antiserum.

Initially, we evaluated, in addition to the sc-695 antibodies, three antisera (ab27497; Abcam, Cambridge, MA; directed toward N-terminal amino acids 2/14; ab10653; Abcam; and E 4644; Sigma-Aldrich, St Louis, MO, both raised against the EpoR extracellular domain). While ab27497 did not work, both ab10653 and E 4644 identified only a single band of 65 to 66 kDa (the molecular mass of EpoR) in UT-7/Epo, a cell line overexpressing EpoR. Because the signal obtained with ab10653 was remarkably stronger, we used this antiserum in all the subsequent studies (Fig 1A).

The 65 to 66 kDa band was absent in K562 cells, which are known to express the receptor at very faint level (Fig 1A).³ When the same samples were analyzed by the sc-695 antiserum, we identified only one signal that occurred in UT-7/Epo cells and not in K562. This signal corresponded strictly to the band detected by the ab10653 antibodies (Fig 1A). The sc-695 antiserum immunoblotting also shows a strong band at 70 to 71 kDa that was also evidenced by a mouse monoclonal raised against HSP70 (Santa Cruz sc-24).

Then, we prepared nuclei, cytosol, and membrane fractions from UT-7/Epo cells (Fig 1B). By analyzing these samples using the sc-695 antiserum, we observed that the putative HSP70 signal was detectable mostly in the cytosol and in part in the membrane. Conversely, the 65 to 66 kDa signal was detectable only in the membrane fraction. The analysis of the same filter by ab10653 evidentiated a signal at 66 kDa only in the membrane fraction. Finally, a third analysis of same blot by sc-24 antibodies identified only the 70 kDa band.

Subsequently, we precipitated extracts of UT-7/Epo and K562 cells by the sc-695 antiserum and blotted the immunoprecipitated materials with the ab10653 antiserum (Fig 1C). We evidentiated only a strong signal a 65-66 kDa in the immunoprecipitated materials from UT7/Epo. When the blot was incubated with the sc-695 antiserum, we again detected only a band at 65 to 66 kDa (Fig 1D). Accordingly, when the sc-695 immunoprecipitate was tested with the sc-24 mAb, we observed that HSP70 remained in the supernatant (results not reported).

The data reported demonstrate the scarce specificity of sc-695 antisera and support the view that this antiserum mostly recognizes HSP70 protein. It has been definitely demonstrated that HSP70 levels increase in cancer and that its build-up is generally associated with a poor prognosis.⁴ Thus, the occurrence of high HSP70 content in tumor samples is not only an expected finding but also a strong confounding factor when the sc-695 antiserum is employed for the detection of EpoR.

In conclusion, because the sc-695 antiserum recognizes proteins other than EpoR, the data in the article by Henke et al need to be interpreted with caution and their observations should be confirmed with more specific anti-EpoR antibodies.

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

The authors indicated no potential conflicts of interest.