Time for revival of the red blood cell count and red cell mass in the differential diagnosis between essential thrombocythemia and polycythemia vera?

Red blood cell indices, red blood cell mass and bone marrow biopsy in the differential diagnosis between essential thrombocythemia and polycythemia vera?

The correct diagnostic classification of the Philadelphia-negative chronic myeloproliferative neoplasms (MPN) in the three subcategories, essential thrombocythemia (ET), polycythemia vera (PV) and primary myelofibrosis (PMF), relies upon diagnostic criteria that aim at minimizing misclassification.1 Several reports have addressed the issue that JAK2V617F positive “ET” patients are frequently misclassified since they actually have a diagnosis of PV.82 This misclassification is partly based upon the use of the hemoglobin (Hb) concentration as a surrogate marker for the red cell mass (RCM), irrespective of the fact that the Hb concentration is influenced by iron deficiency, which is prevalent in PV patients. Indeed, these concerns have been addressed and confirmed in several studies showing that a high proportion of ET patients (approx. 45-65%) did not meet the World Health Organization (WHO) diagnostic criterion of an elevated Hb, despite an increased Cr-51 RCM.75 However, despite the convincing data published in 2005 by Johansson et al.,5 these concerns were not translated into the revised 2007 WHO diagnostic recommendations. These recommendations, therefore, remained unchanged9 and were addressed and met by alternative diagnostic approaches in ET, PV and PMF patients.2 In their 2013 study,3 Silver et al. for the first time prospectively evaluated the accuracy of the 2007 WHO criteria for diagnosing PV, especially in “early-stage” patients. This and other studies support the latest updated WHO criteria (2016) for diagnosing MPN,91 which included these novel data with regard to the inaccuracy of the Hb-concentration, and even the hematocrit (HCT), in the differential diagnosis between ET and PV patients by lowering the Hb/HCT thresholds (> 16.5 g/dL/0.49 in men and > 16 g/dL/0.48 in women).1 Thus, the 2013 Silver study in a prospective setting and with a median 5-year follow up time convincingly demonstrated that the surrogate markers Hb and HCT are inadequate in the assessment of an increased RCM for early PV cases, since 64.3%, 28.5%, and 28.5% of their patients would not have been diagnosed as PV using Hb, HCT, and either Hb or HCT values, respectively.3 Importantly, of the 28 patients with an increased RCM in their study, 18 did not meet the WHO 2007 criteria for an increased Hb value. For the four women, the median Hb count was 15.2 g/dL (range: 14.4-16.4 g/dL) and for the 14 men 17.2 g/dL (range: 15.6-18.1 g/dL), respectively. Similarly, eight patients (1 woman and 7 men) did not meet the WHO criteria for an increased HCT value, being 44.3% for the woman, and for the seven men the median HCT was 48.5% (range: 45.7-49.4 %).3 Silver et al. also highlighted the value of bone marrow (BM) morphology3 as emphasized in the WHO classification. Accordingly, this study supported previous reports by Johansson et al.,5 Cassinat et al.,6 and Alvarez-Larran et al.7 which all revived ancient knowledge, written by the Polycythemia Vera Study Group (PVSG), and underscoring the inaccuracy of the Hb and the HCT values for diagnosing PV and the need for RCM measurement instead.10 This has since fostered intense debate in several reviews and perspective papers expressing conflicting opinions. On the one hand, some authors believe that only RCM measurements can reliably distinguish PV from other MPN,1142 while others would disregard RCM measurements,1512 arguing that Hb/HCT thresholds should be used as surrogate markers for RCM measurements. This lively debate has recently been further fueled by a comprehensive and scholarly review on MPN, emphasizing the urgent need for RCM investigations to distinguish PV from other MPN,16 adding that BM morphology has no place in the distinction of PV from other MPN subtypes.16 Others have highlighted the importance of BM morphology in PV and its usefulness in distinguishing between ET and PV.21173 A very recent study has established a clear-cut distinction between ET and PV, and, therefore, also the reproducibility of BM morphology in so-called masked polycythemia vera (mPV) and its differentiation from ET.21 The disease entity mPV will be further addressed below.

Do the revised and lowered thresholds for Hb/HCT levels unmask undiagnosed PV patients in the general population when “potential PV patients” are being referred?

In 2014, the issue as to which of the three red cell parameters, Hb, HCT or RCM, to use as the diagnostic hallmark of PV was thoroughly reviewed by Barbui et al. They also critically addressed the validity and applicability of the three major diagnostic classification systems for PV as proposed by the PVSG, the British Committee for Standards in Hematology (BCSH), and the WHO.13 It was suggested either to reduce the thresholds for the Hb-concentration, or to include the HCT as a major diagnostic criterion in association with the JAK2V617F mutation. In this review, the existence of prodromal (latent PV-disease), named “masked” PV and defined by Barbui et al.,13 was also addressed, covering patients not meeting the required Hb or HCT threshold levels as defined in the WHO and BCSH criteria.2322 With regards to the impact of lowering the Hb and the HCT thresholds for PV, influenced by the above studies,2322853 the 2016 revised WHO criteria have not resolved the conflicting opinions,2524 as recently addressed and discussed in depth.2726 However, very interestingly, applying the lower Hb thresholds as reported by the WHO 2016 criteria in the Canadian population, Ethier et al. found an Hb value at or above the threshold in 4.1% of all complete blood counts from unselected males and in 0.35% of females. These figures increased the incidence of “potential PV patients” by up to 12-fold in males and 3-fold in females. The same pattern was demonstrated when including the neutrophil and platelet count, implying that up to 60 times more males and three times more women would be suspected of suffering from MPN and would accordingly require diagnostic investigations.24 According to the screening procedure in clinical practice as described by Rumi and Cazzola,25 the best compromise between the need for an early diagnosis of PV patients and the risk of excessively expanding the number of potential PV patients would be a threshold of 17 g/dL in men. Applying this Hb cut-off value, Barbui et al. found that 14% of their 375 patients presenting with WHO-defined PV did not meet the 2016 criteria.26 In the context of screening procedures in clinical practice and which blood cell counts to use, according to the Canadian data, the Hb/HCT thresholds as defined in the 2016 WHO criteria will markedly increase the number of individuals with suspected PV in the general population.24 Given this, it is important to note that a very recent Danish study has found MPN to be massively underdiagnosed with an estimate of 10,000 undiagnosed MPN in Denmark, corresponding to approximately 550,000 US citizens having an undiagnosed MPN and accordingly being at a considerable risk of thrombosis.28 In this perspective, it might be much more cost-effective to screen high-risk MPN individuals and obtain a diagnosis earlier rather than later when the individual has already suffered one or more potentially life-invalidating thrombotic events before being diagnosed with ET or PV.28

Consequences of misclassification of PV as ET

There are several consequences of misclassifying PV patients as ET.

• Misclassification of JAK2V617F positive ET patients as ET instead of PV implies that these patients are not phlebotomized and are, therefore, exposed to an increased risk of potentially life-invalidating or life-threatening thrombotic complications due to the increased RCM. This is not a trivial risk, in particular in those patients aged <60 years without a prior thrombosis but with leukocytosis and platelet counts < 1500 × 10⁹/L since these patients are categorized as “low-risk” according to international guidelines and are not offered cytoreductive treatment, irrespective of the fact that both leukocytes and platelets are deeply involved in the development of atherothrombosis,^29–33 the JAK2V617F mutation promotes atherothrombosis, being associated with e.g. transitory cerebral ischemia, completed stroke and ischemic heart disease, and leukocytosis per se is considered a risk factor for thrombosis in the background population,³⁴ and a causative factor for thrombosis in ET and PV patients.³⁵ Importantly, the most recent studies, including a meta analysis study, have provided evidence that leukocytosis is a risk factor for thrombosis in the MPN-population as well.^36,37
• Misclassifying JAK2V617F positive “ET” patients as “ET” instead of PV has a huge impact on any prognostic model that compares the prognosis of ET and PV patients, the potential outcome being that JAK2V617F positive ET patients have an inferior prognosis as compared to those who are JAK2V617F negative or CALR-positive.³⁸ In several published studies these differences might be explained by the fact that PV patients not being phlebotomized have been included in the ET-cohorts. This may also hold true for mPV patients who have an increased risk of thrombosis (young patients)³⁹ and poorer survival than PV patients. This is likely explained by the fact that several mPV patients have not been phlebotomized despite an expanded RCM.
• Results from studies on safety and efficacy of any drug, both those conventionally used [e.g. hydroxyurea (HU), interferon-α2 (IFN) and anagrelide] and novel agents such as ruxolitinib or experimental drugs (in clinical trials for future approval for the indication of ET or PV) may be severely undermined, impossible to interpret, and therefore not credible.
• Building future therapeutic recommendations and prognostic models on a diagnostic platform that does not take into account the true nature of a disease (e.g. a higher rate of thrombosis in PV than ET, a higher rate of transformation to myelofibrosis and acute myelogenous leukemia in PV than in ET, a reduced life expectancy in PV as compared to ET) due to diagnostic misclassification undermines our current understanding and concepts on MPN in highly important issues. These include pathogenetic mechanisms for disease evolution both in terms of molecular phenotypes, clinical phenotypes and associations between them, diagnostic classification in the biological continuum from early cancer stages (ET, PV) to the advanced myelofibrosis stage, and, not least, when and how to treat MPN. The cornerstone treatment of PV is phlebotomies, carried out to alleviate the hyperviscosity state due to an expanded RCM and thereby to reduce the risk of the deadly thrombosis seen in median survival figures for PV patients of 18 months without such treatment. This approach may, however, be misguided by using only the Hb concentration and the HCT as these are profoundly influenced by the iron-deficient state in patients with PV and also in JAK2V617F positive ET patients in whom erythropoietin (Epo) and ferritin levels and the mean corpuscular volume (MCV) values have been repeatedly reported to be lower than in JAK2V617F negative ET patients.⁴⁰
• Without an estimation of RCM and plasma volume in the diagnostic setting of JAK2V617F positive ET and PV, the transitional stage between ET and PV may be wrongly described as a new disease entity within MPN.²² Thus, it is tempting to speculate whether the “novel” disease entity (mPV) would ever have been born, if arguing that a large proportion of these mPV patients are only “masked” as long as RCM is not being estimated. As discussed above, reports on masked PV^22,23 were influential in lowering the Hb/HCT thresholds in the 2016 WHO classification of MPN.^1,27 This was defined as a new JAK2V617F-positive entity with a phenotype mimicking ET (isolated thrombocytosis) but, as in PV, associated to endogenous erythroid colony formation (EEC) or the BM features of PV, which had previously been described as latent or inapparent PV.^41,42 The revised WHO 2016 classification was, among others, based upon the mPV studies, which defined threshold values as optimal cut-off levels for distinguishing JAK2V617F ET from mPV (Hb 16.5 g/dL/HCT 0.49 in men and 16 g/dL/HCT 48 % in women, respectively)⁴³ and were subsequently validated in larger cohort studies.^15,44

Very recently, the notion that mPV may not be a novel disease entity but has emerged consequent to the inaccuracy of diagnosing PV in the absence of an estimation of RCM has been supported by a large French single center study of 2,480 RCM estimations in patients with JAK2V617F positive ET, “masked PV”, and PV.45 This study showed that patients with mPV actually have an increased RCM and are, therefore, easily "unmasked and revealed to be true PV once RCM is estimated.45 Thus, these mPV patients share clinical and biological features with both ET and PV, with a median age, platelet, Hb and leukocyte levels comparable to those of ET patients, and at the same time classic PV features, (which, in addition to the increased RCM, also include lower Epo level and lower MCV), and have splenomegaly more frequently than ET patients.45 Accordingly, patients being described as mPV nicely present a picture of a diagnosis of MPN as a moving target that is highly dependent on the time point for diagnosis in the biological continuum from early JAK2V617F positive ET to overt PV.

The mPV story underscores the urgent need for a renaissance of the RCM and plasma volume assessment in these patients, since otherwise JAK2V617F positive “ET” patients and mPV patients will not receive adequate treatment by phlebotomies and so will obviously also have an increased risk of thrombosis.3938 The French and other studies, including those pioneered by Silver and Spivak, also put into perspective the view that it is indeed possible to incorporate RCM estimations into ‘good clinical practice’ in the differential diagnosis between JAK2V617F positive ET and PV.451682 However, as noted above, there are still conflicting opinions as to the need for RCM measurements in distinguishing between patients with JAK2V617F positive ET, mPV and overt PV.151142 Of note, a very recent study showed that when applying the 2016 WHO criteria, increased RCM was significantly associated with increased Hb/HCT (93.8%/94.6%),15 thus supporting the 2016 WHO criteria for PV, implying Hb/HCT values should be used as surrogate markers for RCM measurements.15 In this study, the importance of BM morphology for a diagnosis of PV was also highlighted.15

Red blood cell count as a surrogate marker for red cell mass?

Recently, Michiels et al. underscored the importance of RBC count in addition to a BM biopsy as a powerful tool to differentiate between ET and PV.8 In their study, the diagnostic value of RCM in relation to RBC count, Hb and HCT in discriminating between JAK2V617F ET and PV was assessed. The best correlation was found between RBC count and RCM. Thus, at RCM above 30 mL/kg the RBC count was above 5.8×10/L, and this diagnosed PV in all their patients. All JAK2V617F ET patients had a normal RCM and a RBC count below 5.8×10/L. It was concluded that a RBC count within the normal range (< 5.8x10/L in males and < 5.6x10/L in females) enables JAK2V617F ET to be distinguished from prodromal PV and overt PV. Thus, they also concluded that the RBC count and a BM biopsy might obviate the need for RCM measurement.8

Are hemoglobin and hematocrit values imperfect surrogate markers for red cell mass?

In this issue of Haematologica, Silver et al. convincingly confirm the urgent need to investigate JAK2V617F positive ET patients using RCM estimations,46 repeating the important message that a normal Hb or HCT value does not signify a normal RCM in MPN.16111082 Based upon JAK2V617F positivity, chromium-51 RCM, and BM biopsy morphology, 83 and 39 patients were diagnosed with PV and ET, respectively. Chromium-51 RCM separated PV from ET JAK2V617F, whereas red cell values (Hb, HCT, RBC count) overlapped in 25.0-54.7%. The authors concluded that a significant proportion of PV patients may be underdiagnosed by using only red cell values. Of note, using ROC analyses, the authors found threshold values for the Hb HCT coincidentally similar to the WHO 2016 criteria. Furthermore, it was concluded that (without isotope studies) BM biopsies and serum erythropoietin values should become mandatory since they improve diagnostic accuracy. In this perspective, the paper by Silver et al. is highly relevant and timely. It carries a novel approach into the future and will hopefully promote an optimal classification of MPN by a renaissance of the use of RCM estimations (the “gold standard” for discriminating JAK2V617F ET from PV) as a highly important tool to ensure a correct diagnostic classification of MPN. This will be a major scientific step forward in improving good clinical practice in MPN patients. It will reintroduce RCM and plasma volume estimations as essential for correct diagnostic classification of MPN, at least in JAK2V617F positive "ET" patients. In patients with huge splenomegaly but still a low normal or even lowered HCT, in the transitional stages between PV and post-polycythemic myelofibrosis, a RCM estimation may reveal the true nature of the disease as PV and accordingly a need for phlebotomies due to an expanded RCM in patients who still present a normal Hb-concentration and normal HCT consequent to hemodilution due to an expanded plasma volume. As previously noted, RCM estimation in JAK2V617F positive ET patients will likely reduce the risk of thrombosis in a substantial proportion of “ET” patients since they will be correctly classified as PV and will, therefore, receive treatment with phlebotomies. Additionally, these patients will then also have the opportunity to be treated with IFN in cases in which the institution does not include ET patients as an indication for being treated with IFN but does so if the patient has PV. This will be even more important when the new IFN-drug, ropeginterferon α-2b, has been licensed for use in newly diagnosed PV in Europe, and hopefully soon in the US as well.47 Based upon the above considerations, the paper by Silver et al. is also of utmost importance since by diagnosing many more JAK2V617F ET patients correctly as PV, their findings may offer more patients the opportunity to be treated with IFN, which is the only agent that has shown to be disease modifying. In fact, in a subset of patients with early MPN disease (ET and PV), after approximately five years of IFN therapy it was seen that this approach may induce minimal residual disease (MRD), with normal cell counts, normal spleen size, a normal BM, and no detectable JAK2V617F mutation, representing an MRD stage that may even be sustained after interrupting IFN for up to three years.48

Considering the findings by Silver et al. and Michiels et al., in some of their previous papers the RBC count is the most valuable parameter and is better than Hb-concentration and hematocrit when considering the equation: HCT= RBC count x MCV, and taking into account that several PV patients have lowered MCV which accordingly lowers the HCT and wrongly dictates that a phlebotomy is not needed, irrespective of the fact that the RBC count is increased49 (Figure 1). In this context, it is also important to note that erythrocytosis in PV usually induces plasma volume expansion50164 which may mask the true HCT, implying that the HCT in many PV patients, especially women, appears to be normal.50411642

Figure 1.Association between the hemoglobin (Hb)-concentration, hematocrit, red blood cell (RBC) count and mean corpuscular volume (MCV) values in a 76-year old woman with polycythemia vera. The Figure illustrates that: (i) the RBC count is a a more accurate indicator of erythrocytosis than the Hb-concentration and the hematocrit; (ii) that this dissociation is consequent to iron deficiency as evidenced by a lowered MCV; and (iii) the hyperviscosity state due to the raised RBC count is reflected in a low erythrocyte sedimentation rate (SR) (< 1 or 2 mm/h) (normal range: 2-20 mm/h). The need for phlebotomies in this patient was monitored by the elevated RBC count and the lowered SR, and tightly associated with the emergence of headache, which immediately resolved after phlebotomy.

These considerations are not only relevant at the time of diagnosis but also during the course of PV when several patients may not be phlebotomized when only using HCT and elevated RBC count is not taken into consideration (Figure 1). On the contrary, hydroxyurea (HU)-treated patients may be unnecessarily phlebotomized due to an HU-induced increase in MCV and accordingly also an increase in the HCT, although the RBC count is normal.

Hopefully, based upon previous reports on the need of the RCM in the diagnosis of MPN,504916111072 and the most recent studies by Silver et al. and the French study discussed above, consideration of the RCM will be revived at many more MPN centers worldwide. Such efforts are not only expected to improve quality of life of the large proportion of undiagnosed PV patients amongst JAK2V617F positive ET patients, but likely prognosis as well, since they will be correctly diagnosed as PV and accordingly receive the cornerstone treatment of PV (phlebotomy) to reduce the HCT <0.42 in women and <0.45 in men.5350161142

The important distinction between different HCT levels for women and men when deciding the need for phlebotomies has been addressed in several papers.53501642 The rationales for this distinction are several and obvious, including the simple fact that women and men have different red blood cell volumes as reflected in different ranges for red cell indices. This common knowledge dictates that a woman’s normal RCM is approximately 600 mL lower than that for men.534 Accordingly, a female patient with a hematocrit of 45% has at least an excess of approximately 600 mL blood4 which associates with an increased risk of major thrombosis.54 Indeed, the study by Marchioli et al. clearly showed that allowing HCT between 0.45 and 0.50 is associated with a significant risk of death from cardiovascular causes or major thrombotic events54 also in the general population.55 These are lessons that we learnt from Pearson 30 years ago5251 and which have been repeated since then in several other studies: an elevated HCT is associated with an increased risk of thrombosis. In fact, in the general population the risk of thrombosis at elevated HCT values has previously been reported to be driven by smoking,56 which has recently been associated with an increased risk of MPN.57

The excess blood volume is even larger in PV-patients with hepatic vein thrombosis,16 who often have a normal HCT due to an expanded plasma volume.4 Importantly, the thrombosis risk in JAK2V617F positive “ET” patients will likely be markedly reduced simply due to normalization of the expanded RCM by phlebotomies. Additionally, without a RCM estimation, some patients with JAK2V617F positive “ET” may be erroneously classified as “early prefibrotic myelofibrosis” while actually having undiagnosed PV for several years and then being referred with an enlarged spleen, a normal Hb-concentration and a normal HCT, red cell values that are in the normal range due to hemodilution consequent to the expanded plasma volume associated with the enlarged spleen. Accordingly, in such patients, a RCM estimation may reveal an expanded RCM requiring phlebotomies to omit thrombotic complications, often at unusual sites such as portal thrombosis, mesenteric thrombosis and thrombosis of hepatic veins.58 Indeed, similar to mPV as a transitional stage in the biological continuum from ET to overt PV, one might speculate as to whether a proportion of JAK2V617F positive patients with a normal Hb/HCT and splenomegaly classified as “early prefibrotic myelofibrosis” may actually have PV with an expanded RCM and expanded plasma volume in a transitional stage towards classic myelofibrosis.

Today, we still need to go over the important lessons from the history of MPN. Back in 1908, Osler taught us that the RBC count is superior to the Hb concentration as an indicator of erythrocytosis.6059501642 This should, therefore, be used in the diagnosis of PV, and the lesson from the PV study group and from several authorities thereafter is that the RCM is “the gold standard” for an accurate diagnosis of PV in patients with mPV and its precursor stage: JAK2V617F positive ET.1682 In the paper by Silver et al. in this issue of Haematologica and in other papers, the importance of these lessons have been repeatedly highlighted. These will hopefully stimulate research into MPN towards additional comparative and correlative studies on the value of RCM estimations, the RBC count, and BM morphology in the diagnosis of MPN. Such studies are even more urgent taking into consideration a most recent review challenging and critically discussing the role of the hematocrit as a determinant risk factor for thrombosis in erythrocytosis.61 It is to be hoped that such studies may promote a consensus amongst MPN experts that the RCM is essential for a correct classification of JAK2V617F positive ET patients, mPV and PV patients.

This will ensure timely treatment with phlebotomies in those patients who otherwise will be classified as ET and who would then carry an increased risk of potentially life-threatening or life-invalidating thrombotic complications. Future studies should also address whether the RBC count in addition to the erythrocyte SR may be simple but highly robust and reproducible indicators of an increased RCM and the hyperviscosity state, respectively, to be used in the diagnosis of PV and when monitoring PV patients for the need for phlebotomy.62