Diagnosis, management, and outcomes of drug-induced erythrocytosis: a systematic review

Abstract

Secondary erythrocytosis refers to an elevation in hemoglobin or hematocrit due to elevated serum erythropoietin levels. Medications including testosterone and sodium-glucose cotransporter-2 (SGLT-2) inhibitors are increasingly recognized as causes of secondary erythrocytosis. We conducted a systematic review to inform the clinical management of drug-induced erythrocytosis. Following Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines, we performed a systematic literature search in MEDLINE, EMBASE, CENTRAL (all via Ovid), and Google Scholar. Of the 2036 articles screened for eligibility, 45 studies were included in our review, with 35 studies on testosterone and other androgen use, 5 studies on SGLT-2 inhibitors, 3 studies on antiangiogenic tyrosine kinase inhibitors (TKIs), 1 study on erythropoiesis-stimulating agents, and 1 study on a treatment regimen for multidrug-resistant tuberculosis. Cisgender and transgender men on prescription testosterone had erythrocytosis rates of up to 66.7%, with intramuscular formulations, higher doses, and older age associated with increased risk of erythrocytosis. Up to 2.7% of men on testosterone therapy developed thromboembolic events. Among individuals on SGLT-2 inhibitors, erythrocytosis rates ranged from 2.1% to 22%, with those who discontinued therapy demonstrating improvement or resolution of erythrocytosis. Thromboembolic events were reported in up to 10% of these individuals. Antiangiogenic TKIs were studied in patients with cancer, with erythrocytosis developing in up to 43.5% of patients. Drug-induced erythrocytosis is a heterogeneous condition for which there is no clear consensus among clinicians about its diagnosis and management. We offer recommendations for clinical practice within the scope of this systematic review, although further research is required.

Introduction

Secondary erythrocytosis refers to an elevation in hemoglobin or hematocrit in response to elevated serum erythropoietin (EPO) levels, rather than intrinsic dysfunction of the bone marrow.¹ The most common cause of secondary erythrocytosis is chronic hypoxia, which can result from conditions such as hypoxic lung disease, obstructive sleep apnea, smoking, and cyanotic heart disease. Other secondary causes include EPO-secreting tumors and posttransplantation erythrocytosis.^1,2

Medications are an increasingly recognized cause of secondary erythrocytosis. Testosterone has been well established in the literature to cause an elevation in hemoglobin and hematocrit.3, 4, 5 More recently, sodium-glucose cotransporter-2 (SGLT-2) inhibitors have also been identified as a cause of drug-induced erythrocytosis.6, 7, 8, 9

Current guidelines on the diagnosis and management of secondary erythrocytosis are limited. In 2005, the British Society of Haematology published guidelines¹⁰ on the investigation and management of erythrocytosis, with a subsequent amendment¹¹ in 2007 to include JAK2 mutation testing due to its role in polycythemia vera, a primary cause of erythrocytosis. These guidelines recommended a thorough history and physical examination, complete blood count (CBC) and peripheral film, JAK2 mutation testing, and other investigations to identify primary vs secondary causes.^10,11 More recently, these guidelines were updated to provide recommendations for the management of several secondary causes of erythrocytosis; however, drug-induced erythrocytosis was not addressed.¹²

Although guidelines for the management of drug-induced erythrocytosis related to testosterone are available, there is variability in the definition of erythrocytosis and recommendations. The 2018 American Urological Association guidelines¹³ recommended withholding testosterone if hematocrit exceeds 50% to investigate for an etiology and reducing the dose or temporarily discontinuing testosterone for hematocrit ≥54%. In comparison, the 2018 Endocrine Society guidelines¹⁴ recommended withholding testosterone for hematocrit >54% until hematocrit normalizes, then resuming therapy at a lower dose; they also included therapeutic phlebotomy as an effective management strategy. Evidence to support specific hematocrit thresholds is lacking, and considerable uncertainty remains surrounding optimal management.¹⁵ Moreover, existing guidelines do not include suggestions on diagnostic investigations for other secondary causes of erythrocytosis and do not mention the role of cytoreductive therapy, antiplatelet agents, or anticoagulation. Furthermore, since the publication of these guidelines, new drugs associated with erythrocytosis have emerged, namely SGLT-2 inhibitors.6, 7, 8, 9

Current guidelines highlight the lack of consensus on the clinical approach to drug-induced erythrocytosis and the need for further clarification of the clinical outcomes in these conditions. Existing literature does not suggest a clear association between drug-induced erythrocytosis and an increased risk of thromboembolic disease, although there is conflicting evidence for testosterone and thromboembolic risk existing independently of testosterone-induced erythrocytosis.16, 17, 18 We, therefore, conducted a systematic review of the literature on drug-induced erythrocytosis. Our primary objective was to evaluate the latest evidence on the diagnosis, management, and clinical outcomes of drug-induced erythrocytosis to inform a clinical approach to diagnosis and management of these conditions. Secondary objectives included the determination of clinically relevant subtypes, investigations for and management of special populations (including pregnancy and perioperative patients), and outcomes (including thrombosis, bleeding, and mortality) of patients with drug-induced erythrocytosis.

Methods

Search strategy

This is a substudy of the systematic review protocol registered with PROSPERO (PROSPERO CRD42024508643).¹⁹ This review and its findings are reported in accordance with the Preferred Reporting Items for Systematic Review and Meta-Analysis-Protocols (PRISMA-P) guidelines.²⁰ A PRISMA-P checklist for this protocol is shown in supplemental Table 1. A systematic search was performed in MEDLINE, EMBASE, the CENTRAL Register of Controlled Trials, and Google Scholar, from the date of inception to 22 January 2024. The bibliographies of relevant systematic reviews were also hand-searched for relevant studies. Information sources were limited to peer-reviewed literature. A sample search strategy has been included in supplemental Table 2, which was adapted for each of the aforementioned sources; this strategy encompassed all etiologies of secondary erythrocytosis.

Study selection

Two reviewers (J.L. and C.C.H.) independently reviewed the titles and abstracts of articles; any disagreements were resolved by a third party (J.H.). Randomized controlled trials and nonrandomized studies were included. Single-case reports and case series with N < 5 were excluded. Studies involving participants aged <18 years and animal studies were excluded. Publication date was restricted to 2005, the year of JAK2 discovery, or later. No restrictions were placed on language. Studies were categorized by etiology of secondary erythrocytosis, and those not relevant to drug-induced erythrocytosis were excluded. COVIDENCE was used to manage search results. For every eligible study, data were extracted by 1 author (J.L.) using an Excel database to manage the extracted data. An expert librarian with expertise in COVIDENCE, literature searches, and data extraction managed and coordinated this process. Data collection forms are available upon request.

Risk of bias

The Cochrane Handbook risk of bias tool was used to assess randomized controlled trials.²¹ Case-controlled studies were assessed using the Newcastle-Ottawa scale.²² Two reviewers (J.L. and C.C.H.) independently assessed studies for risk of bias; any disagreements were resolved by a third party (B.C.-Y.). A summary of the assessment of bias for individual studies is included in Table 1.

Table 1.

Characteristics of the included studies by type of drug-induced erythrocytosis

Study	Location	Sample size	Study design	Type of drug	Risk of bias assessment
Testosterone in cisgender individuals (prescription)
Abildgaard et al23	Denmark	N = 126	Retrospective cohort study	Testosterone undecanoate (intramuscular)	Good
Bachman et al24	United States	N = 109	Secondary analysis of randomized study	Testosterone enanthate (intramuscular)	Good
Bachman et al25	United States	N = 166	Randomized controlled trial	Testosterone gel	Good
Best et al26	United States	N = 60	Retrospective cohort study	Testosterone cypionate (intramuscular) Intranasal testosterone gel	Good
Conaglen et al27	New Zealand	N = 179	Retrospective cohort study	Testosterone undecanoate (intramuscular) Testosterone pellet	Poor
Coviello et al28	United States	N = 121	Secondary analysis of randomized study	Testosterone enanthate (intramuscular)	Fair
Du Plessis et al29	South Africa	Not reported	Retrospective cohort study	Testosterone undecanoate (intramuscular)	Poor
El-Khatib et al30	United States	N = 263	Retrospective cohort study	Testosterone cypionate (intramuscular)	Good
Hayden et al31	United States	N = 97	Retrospective cohort study	Testosterone pellet	Good
Ip et al32	Australia	N = 158	Retrospective cohort study	Testosterone pellet	Poor
Jick et al33	United Kingdom	N = 5 841	Prospective cohort study	Testosterone undecanoate (intramuscular)	Good
Kavoussi et al34	United States	N = 1180	Retrospective cohort study	Intramuscular testosterone Transdermal testosterone gel Testosterone pellet Clomiphene citrate	Poor
Levcikova et al35	Slovakia	N = 69	Cross-sectional study	Testosterone undecanoate (intramuscular)	Poor
McClintock et al36	United States	N = 26 586	Retrospective cohort study	Intramuscular testosterone (multiple) Testosterone pellet Topical testosterone	Good
Ory et al37	United States	N = 48 671	Retrospective cohort study	Not reported	Good
Pastuszak et al38	United States	N = 178	Retrospective cohort study	Testosterone cypionate (intramuscular) Testosterone pellet Testosterone topical gel	Fair
Reddy et al39	United States	N = 78	Retrospective cohort study	Testosterone cypionate (intramuscular) Testosterone pellet Intranasal testosterone	Good
Rivero et al40	United States	N = 81	Randomized controlled trial	Testosterone cypionate (intramuscular) Intranasal testosterone gel	Poor
Rotker et al41	United States	N = 228	Retrospective cohort study	Testosterone pellet	Poor
Sinclair et al42	Australia	N = 101	Randomized controlled trial	Testosterone undecanoate (intramuscular)	Good
Van Buren et al43	United States	N = 622; therapeutic phlebotomies in 2014 (n = 193), 2015 (n = 212), and 2016 (n = 239)	Retrospective cohort study	Not reported	Poor
Vorkas et al44	United States	N = 21	Case control study	Not reported	Fair
Wheeler et al45	United States	N = 363	Retrospective cohort study	Intramuscular testosterone Transdermal testosterone gel Testosterone pellet Clomiphene citrate	Good
Testosterone in transgender individuals (prescription)
Antun et al46	United States	N = 983	Prospective cohort study	Not reported	Good
Cheng et al47	United States	N = 30	Retrospective cohort study	Testosterone pellet	Poor
Madsen et al5	Amsterdam	N = 1073	Retrospective cohort study	Testosterone undecanoate (intramuscular) Testosterone ester mix Intranasal testosterone	Fair
Nolan et al48	Australia	N = 180	Retrospective cross-sectional study	Testosterone enanthate (intramuscular) Testosterone undecanoate (intramuscular)	Good
Nolan et al49	Australia	N = 67	Retrospective cross-sectional study	Testosterone gel	Poor
Nolan et al50	Australia	N = 72	Retrospective cohort study	Testosterone cream	Poor
Nolan et al51	Australia	N = 64	Randomized controlled trial	Testosterone undecanoate (intramuscular) Testosterone gel	Poor
Oakes et al52	United States	N = 519	Retrospective cohort study	Testosterone undecanoate (intramuscular) Transdermal testosterone	Fair
Porat et al53	United States	N = 282	Retrospective cohort study	Not reported	Poor
Tatarian et al54	United States	N = 511	Retrospective cohort study	Testosterone cypionate (intramuscular) Testosterone undecanoate (intramuscular)	Fair
Testosterone in cisgender and transgender individuals (prescription)
Middleton et al55	Australia	N = 347; cisgender (n = 251) and transgender (n = 96)	Prospective cohort study	Testosterone undecanoate (intramuscular)	Poor
Androgen abuse (nonprescription)
Smit et al56	The Netherlands	N = 100	Prospective cohort study	Not reported	Poor
SGLT-2 inhibitors
Chin-Yee et al7	Canada	N = 891	Retrospective cohort study	Not reported	Poor
Gangat et al8	United States	N = 100	Retrospective cohort study	Canagliflozin Dapagliflozin Empagliflozin	Poor
Gill et al57	Canada	N = 42	Retrospective cohort study	Canagliflozin Dapagliflozin Empagliflozin	Poor
Lassen et al58	Denmark	N = 7 926 (SGLT2i group), N = 13 765 (DPP4i group)	Retrospective cohort study	Not reported	Good
Liu et al9	Canada	N = 82	Retrospective cohort study	Canagliflozin Dapagliflozin Empagliflozin	Poor
TKIs and other anti-angiogenic medications
Alexandrescu et al59	United States	N = 5	Case series	Sunitinib Sorafenib	Not applicable
Legros et al60	France	N = 23	Retrospective cohort study	Lenvatinib	Poor
Wang et al61	United States	N = 5	Case series	Axitinib Pazopanib Sorafenib Bevacizumab	Not applicable
EPO
Schreiber et al62	Denmark	N = 52	Randomized controlled trial	EPO beta	Good
MDR-TB treatment
Tesfamariam et al63	Eritrea	N = 219	Retrospective cohort	Kanamycin and/or capreomycin, levofloxacin, ethionamide, cycloserine, para-aminosalicylic acid or pyrazinamide	Fair

Data synthesis

Data were synthesized using descriptive analysis.

Results

Figure 1 shows the PRISMA diagram for study selection. A total of 45 studies were included. Thirty-five studies discussed testosterone: 23 studies examined prescription testosterone in cisgender individuals, 10 studies examined prescription testosterone in transgender individuals, 1 study examined prescription testosterone in both cisgender and transgender individuals, and 1 study examined nonprescription androgen use. Five studies discussed SGLT-2 inhibitors. Three studies discussed tyrosine kinase inhibitors (TKIs) and other antiangiogenic medications. One study discussed EPO. One study discussed an antimicrobial regimen used to treat multidrug-resistant tuberculosis (MDR-TB). The sample sizes of the included studies ranged from 5 to 48 671. Most of the study designs were retrospective (n = 40), but some were randomized controlled trials (n = 5). The median duration of follow-up was 22 months. Serum EPO levels were measured in 7 studies (15.9%). JAK2 mutation testing was ordered in 5 studies (11.1%). Two studies (4.5%) investigated patients for obstructive sleep apnea as an alternative cause of erythrocytosis; otherwise, investigations for secondary erythrocytosis were not reported. Table 1 lists the characteristics of the 45 included studies.

Figure 1.

PRISMA diagram of study selection.

Prescription testosterone for cisgender individuals

Prescription testosterone use in cisgender males for the treatment of hypogonadism was the most common type of drug-induced erythrocytosis in this review, although the clinical definition of hypogonadism varied among studies. A variety of formulations were described, including intramuscular (testosterone cypionate, testosterone undecanoate, testosterone enanthate, and testosterone ester mix), pellet, and topical (transdermal gel, transdermal cream, and intranasal gel). Rates of erythrocytosis ranged from 0% to 66.7%. Intramuscular formulations were associated with higher rates of erythrocytosis and higher levels of hemoglobin and hematocrit than pellet and topical formulations.^26,38, 39, 40^,44,45 Dose dependency was shown, with higher doses of testosterone leading to higher levels of hemoglobin and hematocrit.^28,30 Four studies noted that older men were at a higher risk of developing erythrocytosis on testosterone and tended to have a greater degree of erythrocytosis.^28,24,32,33 The mechanism of this effect has been theorized to be due to age-dependent hepcidin suppression²⁴ or the inhibition of erythropoietic inhibitor production.²⁸

One retrospective study³⁷ described an increased risk of major adverse cardiac events and venous thromboembolism in those receiving testosterone therapy who developed erythrocytosis compared with those who did not (odds ratio, 1.35). Another study³⁴ reported that 10 of 1180 men (0.8%) were diagnosed with deep vein thrombosis (DVT) while on testosterone therapy. However, most of these men (n = 7) had other identifiable risk factors for DVT including postoperative (n = 3), long-distance travel (n = 3), factor V Leiden deficiency (n = 1), and Klinefelter syndrome (n = 1). None of the men had erythrocytosis at the time of DVT diagnosis. No other studies described thromboembolic events.

Three studies^28,24,25 reported the discontinuation of testosterone therapy in those with erythrocytosis for safety reasons, including hematocrit >0.54.^28,24 In 1 study, hemoglobin and hematocrit levels returned to normal 3 months after discontinuation of testosterone.²⁵ Outcomes after discontinuation were otherwise not reported. Other studies implemented dose reduction^27,31 and/or phlebotomy^40,27,41,43 as management strategies for testosterone-associated erythrocytosis in cisgender males. However, the laboratory thresholds to initiate these strategies varied or were not reported; for example, Rotker et al⁴¹ recommended phlebotomy to patients with hematocrit >50%, whereas Rivero et al⁴⁰ referred a patient for phlebotomy for hematocrit >52%. One study³⁵ also recommended the initiation of aspirin 30 mg by mouth daily for patients with hemoglobin >176 g/L, hematocrit >52%, and erythrocytes >6.0 million per μL.

Prescription testosterone for transgender individuals

Similar to cisgender males, multiple formulations of prescription testosterone were found in the literature for transgender and gender-diverse individuals as part of gender-affirming care. Erythrocytosis typically developed within the first 1 to 3 years of therapy,^5,47,54 with the greatest increase in hematocrit also occurring during the first 1 year in 1 study.⁵ In 1 study that conducted long-term follow up in transmasculine individuals for 20 years, hematocrit continued to increase for the duration of the study, albeit only slightly.⁵ Rates of erythrocytosis were as high as 46.7% in 1 retrospective study on patients initiated on testosterone pellets⁴⁷; however, 36% of patients with erythrocytosis during the study period already had increased hematocrit before starting pellet therapy. In other studies, rates of erythrocytosis ranged from 0% to 22%, depending on formulation. Notably, in all studies, erythrocytosis was defined as hematocrit >50%, according to the Endocrine Society guidelines for transgender males or using the laboratory reference intervals for cisgender males.⁶⁴ Generally, long-acting intramuscular formulations of testosterone were associated with a greater risk for erythrocytosis and a higher elevation in hematocrit than topical formulations.^5,48,52,53

Two retrospective studies reported thromboembolic events.^54,52 In 1 study,⁵² 5 of 519 patients (0.9%) developed thromboembolic disease, including superficial vein thromboses, deep vein thromboses of the lower limb, and ischemic stroke; however, only 4 of these patients met the criteria for erythrocytosis during the study period, and none had erythrocytosis at the time of diagnosis of the thromboembolic event. Another study⁵⁴ reported thrombosis in 3 of 113 patients (2.7%), including 2 incidents of DVTs and 1 patient who had recurrent transient ischemic attacks.

Management of testosterone-associated erythrocytosis in transgender individuals typically consisted of observation or dose reduction of testosterone.^54,52 In 2 studies,^54,49 a small proportion of patients switched formulations, with subsequent improvement in hematocrit. Furthermore, in the study by Tatarian et al,⁵⁴ lifestyle modifications including smoking cessation were implemented in 6% of management strategies, and 2 of 62 patients (3.1%) with hematocrit ≥54% were referred to hematology. Only 1 study⁵² initiated phlebotomy for individuals with hematocrit >50%, with 4.8% of patients meeting the criteria for this.

Nonprescription androgen use

One prospective cohort study⁵⁶ investigated nonprescription use of androgen in a group of 100 men. The exact dose of androgen was unknown and varied among participants based on label information. A mean increase in hemoglobin by 0.53 g/L and a mean increase in hematocrit by 3% were described over the course of the 1-year study period; there was no association between the degree of erythrocytosis and the duration of androgen cycle, nor was there an association between the degree of erythrocytosis and a higher androgen dose. Erythrocytosis resolved ∼3 months after the end of the androgen cycle. No thromboembolic events were described.

SGLT-2 inhibitors

Empagliflozin, canagliflozin, and dapagliflozin were all described in the literature to have an association with erythrocytosis. Rates of erythrocytosis on SGLT-2 inhibitors ranged from 10% to 22%.^57,58 One retrospective cohort study⁵⁸ observed an adjusted mean increase in hemoglobin of 4.3 g/L; other retrospective studies^7,9 described a median increase in hemoglobin ranging from 23 to 25 g/L. Individuals who remained on SGLT-2 inhibitors typically demonstrated a plateau in their hemoglobin levels^7,57; 1 study⁹ reported that most of the patients who remained on SGLT-2 inhibitors demonstrated normalization or return to baseline of their hemoglobin levels. Up to 26% of patients discontinued their SGLT-2 inhibitors, with subsequent improvement or resolution of erythrocytosis.^7,8,57

Two retrospective studies described the management strategies for SGLT-2 inhibitor–associated erythrocytosis. In the study by Gangat et al,⁸ 29% of patients with erythrocytosis underwent phlebotomy, although the laboratory threshold, volume, frequency, and target for phlebotomy were not reported. Sixty-eight percent of patients received antiplatelet agents: most of them (95.5%) were treated with aspirin, whereas the remainder received clopidogrel. Another 16% were on systemic anticoagulation. In the study previously conducted by our group,⁹ 14.6% of patients underwent phlebotomy, 12.2% were treated with aspirin, and 1 patient (1.2%) received hydroxyurea.

Gangat et al⁸ reported thrombotic events in 10 of 100 patients (10%), 70% of which were arterial. There was no correlation between thrombotic risk and levels of hemoglobin and hematocrit at the time of the event; however, peak hemoglobin and hematocrit were associated with increased thrombosis. At the time of the thrombotic event, 6 patients were receiving phlebotomy, 5 patients were taking aspirin, and 3 were on systemic anticoagulation. Although there was no significant difference in thrombosis rates in individuals on antiplatelet vs systemic anticoagulation, arterial thrombosis was found to be more likely in those receiving phlebotomy (5 of 29 patients). Our group⁹ also observed thromboembolic events in 6.1% of patients, including ischemic stroke (n = 3 patients), myocardial infarction with a suspected cardioembolic stroke after catheterization (n = 1 patient), and DVT (n = 1 patient). In 3 patients, these events were attributed to other causes, including severe atherosclerotic disease and inadequate anticoagulation for atrial fibrillation, and all 5 patients had other cardiovascular risk factors. Their hemoglobin and hematocrit levels at the time were not documented. In the existing literature, only 1 patient on an SGLT-2 inhibitor had a thromboembolic event associated with erythrocytosis at the time of the event, consisting of a cerebellar stroke when hemoglobin was 173 g/L and hematocrit was 0.52.⁵⁷

TKIs and other antiangiogenic medications

TKIs and other antiangiogenic medications used in cancer treatment have been reported to be associated with erythrocytosis. This effect has been described for lenvatinib for the treatment of hepatocellular carcinoma; sorafenib for the treatment of hepatocellular carcinoma and melanoma; and sunitinib, axitinib, and pazopanib for the treatment of renal cell carcinoma.59, 60, 61 Bevacizumab, an anti–vascular endothelial growth factor (VEGF) monoclonal antibody, has also been associated with erythrocytosis in the treatment of renal cell carcinoma.⁶¹ Erythrocytosis was reported in up to 43.5% of patients on these medications, developing as early as 1 to 2 weeks after the initiation of treatment in some patients, with a mean increase in hemoglobin ranging from 14.1 to 30.6 g/L.59, 60, 61 Alexandrescu et al⁵⁹ found that serum EPO levels were not suppressed below normal, whereas Wang et al⁶¹ and Legros et al⁶⁰ found serum EPO to be elevated in those for whom it was measured. In the case series by Alexandrescu et al,⁵⁹ JAK2V617F mutation testing was performed and was negative for all patients.

The management of erythrocytosis associated with antiangiogenic medications varied. Legros et al⁶⁰ initiated Aspirin 81 mg for primary prophylaxis and phlebotomy for hematocrit >0.52, the latter of which was required in 8.7% of patients. Wang et al⁶¹ also describe the use of phlebotomy in all patients, with most of them (83.3%) only requiring 1 phlebotomy, although the volume was not reported. In patients who discontinued TKIs, hemoglobin levels decreased, suggesting a reversible effect.^59,60 In the case series by Wang et al,⁶¹ most of the patients required dose reduction of TKIs for erythrocytosis.

EPO-stimulating agents

The use of high-dose EPO in patients with progressive multiple sclerosis was studied in 1 randomized controlled trial⁶² and was found to have no significant effect on the composite outcome of maximum gait distance, hand dexterity, and cognition at 24 weeks. Notably, 38.5% of patients receiving EPO treatment required dose reduction (from 48 000 IU weekly or biweekly to 30 000 IU weekly or biweekly) or dose omission for erythrocytosis. Phlebotomy was initiated for hematocrit levels >50% in men or >48% in women, with 23% of patients in the EPO group requiring phlebotomy. No thromboembolic events or deaths were reported.

MDR-TB treatment

One retrospective cohort study⁶³ conducted at a single center in Eritrea assessed 219 patients admitted for MDR-TB and found an association between erythrocytosis and the treatment regimen used, which consisted of kanamycin and/or capreomycin, levofloxacin, ethionamide, cycloserine, and para-aminosalicylic acid or pyrazinamide. Over the mean follow-up time of 2 years, 31 patients (14.2%) developed erythrocytosis, with a median time of onset of 6 months. The development of erythrocytosis was associated with male sex, baseline body weight, educational level, and duration of treatment. After multivariate analysis, educational level was no longer significant. Phlebotomy was initiated in 2 of 219 patients (0.9%), although the indication for initiation of phlebotomy was not reported. The study did not report on thromboembolic events.

Discussion

To our knowledge, this is the first systematic review of the literature on drug-induced erythrocytosis. Our search identified several classes of medications associated with erythrocytosis, including testosterone in cisgender and transgender patient populations, SGLT-2 inhibitors, and TKIs. In single studies, EPO and one treatment regimen for MDR-TB were also associated with erythrocytosis. Few studies reported on the diagnostic strategies undertaken for patients with drug-induced erythrocytosis. In a small number of studies, serum EPO level and JAK2 mutation testing were ordered. Patients rarely underwent formal investigations for other secondary causes of erythrocytosis, such as sleep apnea or hypoxic lung disease; however, this is likely attributable to the fact that most of the studies were retrospective, and many patients already had established diagnoses of other secondary causes.

The most common medication in this review was testosterone prescribed to cisgender men for the treatment of hypogonadism. Intramuscular route of administration, higher dose, and older age were all risk factors for erythrocytosis; however, thromboembolic complications were rare, with only 1 retrospective study describing an increased risk of major adverse cardiac events and venous thromboembolism in those on testosterone therapy who developed erythrocytosis, compared with those who did not develop erythrocytosis.³⁷ Discontinuation of testosterone appears to allow hemoglobin and hematocrit to normalize. Dose reduction, aspirin, and phlebotomy have also been implemented, with phlebotomy initiated at hematocrit levels of 50% to 52%.^40,27,41,35 This is similar to current Endocrine Society clinical practice guidelines¹⁴ on testosterone therapy in males with hypogonadism, which recommends temporarily holding testosterone if hematocrit levels exceed 54% and resuming therapy at a lower dose once hematocrit normalizes; phlebotomy is also recommended as an effective strategy.

In contrast, prescription testosterone in transgender and gender-diverse individuals was rarely discontinued for erythrocytosis in this review. Instead, these individuals were observed or underwent dose reduction of testosterone; changing route of administration and phlebotomy also appear to be effective strategies, in combination with lifestyle modifications. Although current Endocrine Society guidelines⁶⁴ on testosterone therapy in transgender males recommend monitoring of hematological parameters every 3 months and avoiding erythrocytosis, they do not make recommendations for managing testosterone-induced erythrocytosis. In their longitudinal study, Madsen et al⁵ recommended switching injectable testosterone therapy to a transdermal route of administration and implementing weight loss for body mass index >25 kg/m², smoking cessation, and treatment optimization for chronic lung disease and sleep apnea. Thromboembolic events developed in 0.9% to 2.7% of transgender patients on testosterone therapy; the risk of these events remains unclear in this patient population.

In nonprescription androgen use, erythrocytosis resolved 3 months after the androgen cycle was completed, and no thromboembolic events were documented.⁵⁶ Although previous case reports have described thromboembolic events associated with anabolic androgen use, with 1 study describing a fivefold increased risk of thromboembolic disease in these users,^{29,36,42,50,51,55,}65, 66, 67, 68, 69 further studies are required to clarify this relationship.

SGLT-2 inhibitors are increasingly recognized as a cause of drug-induced erythrocytosis, with canagliflozin, dapagliflozin, and empagliflozin all implicated. Similar to testosterone, discontinuation or dose reduction of this class of medications has been seen to result in improvement or resolution of erythrocytosis, whereas those who continue the medication with no dose adjustments tend to experience a plateau in hemoglobin levels.7, 8, 9^,57 The association between SGLT-2 inhibitor–associated erythrocytosis and the risk of thromboembolic disease is unclear. The rates of thromboembolic events in this review were similar to or higher than those observed in randomized controlled trials.⁷⁰ Gangat et al⁸ reported an association of increased thrombosis with peak hemoglobin and hematocrit levels but no association with erythrocytosis at the time of the event; this study also did not find a difference in thrombosis rates in those on antiplatelet agents vs systemic anticoagulation.

The use of antiangiogenic TKIs and anti-VEGF receptor therapies in the treatment of various malignancies has been associated with erythrocytosis in the literature. Although it has been hypothesized that erythrocytosis is a class effect of anti-VEGF therapies, the mechanism underlying this effect is unclear. Alexandrescu et al⁵⁹ suggest that TKIs cause sensitization of the body to the effects of EPO, with VEGF blockade stimulating hypoxia-induced elements and causing rebound erythrocytosis; Wang et al⁶¹ suggest that anti-VEGF therapy causes the transient release of EPO from malignant cells, because serum EPO levels were shown to be increased in patients receiving these treatments. This effect appears to be reversible, with those who discontinue or reduce the dose of the implicated medication demonstrating a subsequent decrease in hemoglobin, which was observable within 1 month for most patients.^61,60

High-dose recombinant EPO in individuals with progressive multiple sclerosis, although not a typical treatment for this condition, was associated with erythrocytosis and was effectively managed with dose reduction or dose omission, as well as phlebotomy.⁶² An antimicrobial treatment regimen used for MDR-TB (consisting of kanamycin and/or capreomycin, levofloxacin, ethionamide, cycloserine, and para-aminosalicylic acid or pyrazinamide) was also associated with erythrocytosis in a retrospective study at a center in Eritrea⁶³; none of these medications have been reported individually to cause erythrocytosis in the literature. Given that 22.5% of the cases developed erythrocytosis after 1 year of treatment, this was felt by the authors to be less likely related to MDR-TB itself. The authors did not speculate on potential mechanisms. Erythrocytosis associated with TB has been remotely described in pulmonary, renal, splenic, and myocardial TB,71, 72, 73, 74 with mechanisms suggested to be reactive to support immune response⁷¹ or due to an abnormal allergic response.⁷⁴

There are several limitations to this review. Our review was limited to peer-reviewed literature; as such, it is inherently limited by publication bias, and we cannot exclude the possibility of relevant studies in other databases that were not searched. There was significant heterogeneity in the included studies, including the types, formulations, and doses of medications as well as in the study quality and designs, with most of the studies being retrospective in nature. As such, the rates of thromboembolism are difficult to interpret, particularly given the lack of reporting for baseline risk of thromboembolism in these studies. Furthermore, several of the findings discussed in this review were demonstrated in relatively small patient populations and may not be generalizable. Nonetheless, our findings demonstrate the heterogeneous nature of drug-induced erythrocytosis and identify multiple classes of medications that may be implicated in erythrocytosis. Further studies would be helpful to establish a clinical approach to drug-induced erythrocytosis and to clarify clinical outcomes in this condition.

Suggestions for clinical practice

Due to the heterogeneity of these patient populations and of the included studies, there is limited evidence to guide clinical practice for drug-induced erythrocytosis. Based on this systematic review of existing literature, we suggest the following.

General recommendations

All patients starting the medications covered in this review should be counseled on the risks of developing erythrocytosis and associated thrombosis risk, while acknowledging that high-quality evidence for the latter is lacking for most medications. We suggest that physicians monitor patients for drug-induced erythrocytosis with a CBC after initiating treatment, with the frequency of ongoing monitoring depending on the type of medication and its indication. The role of phlebotomy or aspirin is uncertain, and these interventions may not be sufficient to mitigate thrombosis risk. Practitioners should discuss the risks and benefits of each treatment and engage in shared decision-making. If relevant, enrollment in a clinical trial or prospective registry should be considered.

Prescription testosterone in cisgender males

This is the area with the most studies (n = 23), including the largest multicenter retrospective study including 48 671 men.³⁷ Based on this, erythrocytosis rates range from 0% to 66.7%, with higher risk associated with intramuscular formulation, higher doses, and older age. Furthermore, individuals on testosterone with erythrocytosis have a higher risk of thromboembolic events. We suggest that physicians regularly monitor patients starting these medications with a CBC every 3 months, select an alternate medication or nonintramuscular route if feasible, use the lowest possible dose of that medication, and manage any other cardiovascular risk factor.

Prescription testosterone in transgender males

There are 10 studies with erythrocytosis ranging from 0% to 46.7%, occurring within 1 to 3 years. Similar to cisgender males, these individuals are at higher risk of erythrocytosis with the use of long-acting and intramuscular formulations. Similar to prescription testosterone for cisgender males, we suggest monitoring these patients with a CBC every 3 months, selecting an alternate medication or nonintramuscular route if feasible, using the lowest possible dose of that medication, and managing any other cardiovascular risk factors.

SGLT-2 inhibitors

There are 5 studies with incidence of SGLT-2 inhibitor–associated erythrocytosis ranging from 2.1% to 22%, occurring at a median of 2.3 years after exposure to an SGLT-2 inhibitor. Thromboembolic events were reported in 2.4% to 10% of cases, although whether these were attributable to SGLT-2 inhibitor–associated erythrocytosis is uncertain, and baseline risk in study populations is unknown. In patients with severe erythrocytosis, management requires weighing the benefits of the SGLT-2 inhibitor against potential risks of erythrocytosis; input from a multidisciplinary team, including hematology, endocrinology, or other relevant specialties, is warranted. The role of interventions including aspirin and phlebotomy are unknown.

Other medications including nonprescription testosterone, TKIs, EPO, and anti-TB medications

Evidence is limited studies for erythrocytosis associated with these medications. Practitioners should consider applying the general recommendations.

Areas requiring further investigation

It is currently unknown which patients are at higher risk of drug-induced erythrocytosis or require more extensive investigations such as bone marrow biopsy or molecular testing. The development of a predictive model or scoring system may help clarify this. Further studies for prospective validation would also be beneficial to determine thromboembolic risk associated with drug-induced erythrocytosis, to establish whether antiplatelet agents or phlebotomy have a role in mitigating this risk, and to clarify long-term follow-up in this heterogeneous patient population.

Conflict-of-interest disclosure: The authors declare no competing financial interests.