The first case report of EPO-induced PRCA was published in the late 1990s. Reports on the condition reached its peak in 2001 and 2002 [3] and continued to appear intermittently thereafter. During treatment with EPO, a sudden decrease in Hb levels, accompanied by markedly reduced reticulocyte count, suggests the possibility of EPO-induced PRCA [4]. The following criteria need to be met for diagnosis of EPO-induced PRCA: (i) the common causes of PRCA have been excluded; (ii) less than 5% of erythroblasts in the bone marrow with otherwise normal cellularity; and (iii) positive antibodies binding EPO detected by available tests [4].

In the first hospital admission of our patient, we considered his severe anemia a result of severe infection and hyporesponsiveness to rHuEPO. As past case report had shown the efficacy of Roxadustat in treating EPO-hyporesponsive anemia [5], we changed his rHuEPO to Roxadustat and his infection was brought under control quickly. His Hb was stable for quite a few days without RBC transfusion and he was discharged.

But during his second hospital admission, his anemia was more severe. By that time, his chest infection was under control and his CRP had normalized. We excluded other causes such as iron deficiency, on-going or latent infection, tumor, bleeding and hemolysis for his severe anemia. The more common causes of PRCA such as lymphocyte and plasma cell disorders, autoimmune disorders, viral infection, thymoma or other suspected medication were ruled out, too. Hence, we felt the sudden loss of efficacy to rHuEPO and a reduction of reticulocytes < 10,000/μL were highly suggestive of EPO-induced PRCA [6]. Meanwhile, the second BM biopsy showed erythroblasts of 5.2%, dropping further from 7% obtained in the first biopsy. Although the percentage of erythroblasts was just over the 5% cut off for PRCA, as the disease progressed, we speculated the erythroblasts would fall below 5%.

Our patient’s EPO-binding Ab test was negative to our surprise. Our literature search found two other case reports of PRCA induced by EPO therapy without demonstrable EPO antibodies [7, 8]. In our case, we speculated a number of plausible reasons for the absence of proven EPO Abs. The recommended assays for identifying EPO Abs included radioimmunoprecipitation (RIP), ELISA, surface plasmon resonance (BIAcore), and bioassays [9]. Each of these tests came with its own advantages and disadvantages in terms of sensitivity, specificity and accessibility [9]. In China, the available tests are limited and we have only been able to use an ELISA method which is relatively inexpensive and easy to use. EPO is combined with a porous plate and a secondary Ab is used to detect the binding of Abs in patient’s serum [6]. Depending on the washing conditions, Abs may bind non-specifically to the plate, whereas low-affinity Abs may be washed away resulting in low specificity and sensitivity of the conventional ELISA assay [6]. The sensitivity of EPO-Ab test may also be affected by timing of the sampling in relation to cessation of EPO therapy or commencement of Roxadustat. In a case series of 13 patients with EPO-induced PRCA, after discontinuation of EPO therapy, the levels of EPO Abs in the patients declined gradually [10]. It has also been speculated that endogenous EPO upregulated by Roxadustat helps neutralize anti-EPO Abs without boosting the formation of EPO Abs [11]. In our case, the first blood sample for EPO-Abs was taken after the patient’s EPO therapy had been stopped and Roxadustat commenced for a month. We speculated that our negative Ab results might be due to the low sensitivity of the ELISA method and/or diminishing level of EPO-Abs. While we suspected that the cause of PRCA in our case was the result of EPO Abs that was not yet made demonstrable, there remained a slight possibility that the PRCA had been due to another cause responding to cyclosporine [12].

Regarding therapy, once EPO-induced PRCA is diagnosed, merely stopping the EPO is often not enough for the recovery of red blood cell production. Such recovery has been reported in patients after treatment with immunosuppressive therapy or renal transplant [13]. Both glucocorticoid and cyclosporine have been shown to be effective in treating renal patients with EPO-induced PRCA [12]. In this case, we have managed to use cyclosporine alone and observed a gradual recovery of reticulocyte counts. We have chosen to avoid steroid to reduce hyperglycemia and the regimen has worked.

In recent years, Roxadustat, an oral HIF-PHI, has demonstrated efficacy and safety in the treatment of kidney disease patients with renal anemia whether on dialysis or not [14, 15]. In our literature search, we found a case of dialysis patient whose EPO-induced PRCA resolved after 3 months of treatment with steroids and cyclosporine, followed by Roxadustat [16]. In another case, the anemia responded to Roxadustat after 5 months of cyclosporine therapy [17].

It is likely that the erythropoietic response to Roxadustat after cyclosporine is related to the mechanistic action of Roxadustat. Roxadustat increases endogenous EPO levels by stabilizing HIF-α that trigger EPO genes through HIF-PHI. However, in patients with EPO-induced PRCA and high Ab levels, we speculate that Roxadustat on its own may be ineffective early on due to the high level of anti-EPO Abs interacting with both endogenous and exogenous EPO [16]. In these cases, only after adequate immunosuppressive treatment and reduction of Ab titers can Roxadustat promote haematopoiesis and alleviate the anemia. It is not clear whether the patient may be safely re-challenged with EPO or he may suffer a relapse [18]. The option of using a HIF-PHI is a challenge and trial. In our case, we chose to use Roxadustat after 2 months of cyclosporine. We have used cyclosporine 250 mg/day in two divided doses to aim to achieve a trough level of 100–150 ng/mL. Steroid has been avoided to reduce hyperglycemia. As the reticulocyte counts rose over the next couple of months, Roxadustat was introduced to stimulate and maintain the erythropoiesis of the patient.

In conclusion, we could say that we have encountered and treated successfully a HD patient who developed PRCA whilst on EPO therapy with negative Ab under an ELISA test. The clinical course, markedly reduced reticulocyte count < 10,000/μL, the BM biopsy revealing reduced erythroblasts, and its subsequent response to cyclosporine, were in keeping with EPO-induced PRCA. Anti-EPO Ab assay is not always positive in EPO-induced PRCA and to which the clinicians should be alerted. With or without demonstrable EPO Abs, our case showed that treatment with cyclosporine and a switch to Roxadustat might be proven and effective in treating this type of severe anemia in patients treated with EPO previously.