Reproducibility of the WHO Histological Criteria for the Diagnosis of Philadelphia Chromosome-Negative Myeloproliferative Neoplasms

Umberto Gianelli; Anna Bossi; Ivan Cortinovis; Elena Sabattini; Claudio Tripodo; Emanuela Boveri; Alessia Moro; Riccardo Valli; Maurilio Ponzoni; Ada M Florena; Giulio F Orcioni; Stefano Ascani; Emanuela Bonoldi; Alessandra Iurlo; Luigi Gugliotta; Vito Franco


Mod Pathol. 2014;27(6):814-822. 

In This Article


The classification of Philadelphia chromosome-negative myeloproliferative neoplasms is a challenging task because of the potential clinical and pathogenic overlap at onset among the different diseases, which clearly reflects on the lack of pathognomonic morphological features; equally important, most bone marrow biopsies are presently performed earlier than in the past in almost 'asymptomatic' patients, whose abnormalities are mostly restricted to peripheral blood count values, such as thrombocytosis.[15–18] In this study, aimed to evaluate the inter-observer reproducibility of the morphological parameters considered for the diagnosis of Philadelphia chromosome-negative myeloproliferative neoplasms, we found an overall high level of agreement (>70%) for about all these features. Interestingly, the percentage of agreement ranged among reviewers in relation to the individual pathological entity and to the different categories of each morphological feature. It is important to note that our study is the only one which analyzes the inter-observer reproducibility of the morphological criteria of the WHO classification considering all the three Philadelphia chromosome-negative myeloproliferative neoplasms. We think that this approach could better reproduce the daily clinical practice.

WHO classification of myeloproliferative neoplasms has been considered as poorly reproducible by some groups, while others managed to reach satisfactory interobservers' agreement; even in this latter event, a variable degree of morphologically unclassifiable cases were always recorded.[8,9,17–19]

This issue is not trivial, especially considering some recent reports, in which the prognostic impact of histology is reached when a central pathology review is asked to experts such as the authors of the WHO chapter on MPN classification. [20–23]

Although the current classification per se enabled to reach inconceivable achievements, the above-reported controversies disclose some drawbacks of this classification scheme of MPN.

In the present study, the lowest scores were obtained when reviewers considered myeloid-to-erythroid ratio (70%) and the presence of naked nuclei (65%), with the former reaching the highest discordance in polycythaemia vera cases. Discordance in the evaluation of the myeloid-to-erythroid ratio, particularly in PV cases, can be understood as this ratio can vary from either normal, if the two lineages increase consistently, or decreased if erythropoiesis is more increased than granulopoiesis. However, as it is relevant for a diagnosis, the use of myeloid- and erythroid-associated markers could be suggested, either always or only in cases of uncertainty. Alternatively, the simultaneous evaluation of bone marrow biopsy and aspirate smears can allow an accurate estimation of the erythroid and myeloid series and their quantification. Moreover, some degree of discordance, and particularly about erythropoiesis, may be attributed to the variable quality of different staining. To this regard, here we like to underline that optimally treated bone marrow biopsies and good quality slides and staining are mandatory prerequisites for the best possible diagnoses and that Giemsa staining can help in the evaluation of erythropoiesis and granulopoiesis. At a variance with the other groups,[18] the concordance about the 'loose cluster' parameter is very high (95%), but it did not bear significance in differentiating the three myeloproliferative neoplasms.

Moreover, we selected 11 out of 18 morphological features that resulted the most statistically useful for the differential diagnosis among the three myeloproliferative neoplasms. Our analysis demonstrated that these selected parameters allow to correctly classify 72% of the cases by morphology alone. Interestingly, primary myelofibrosis is the disease for which the highest level (89%) of consensus has been reached.

To further investigate the relationship between morphological profiles identified by four reviewers and 'consensus' diagnosis, we performed a multiple correspondence analysis on the 11 selected variables. Analysis of the results in Figure 2 can help to identify the more useful parameters in the diagnosis of a myeloproliferative neoplasms; (a) cellularity: this parameter, if normal for the patient's age, seems crucial in the diagnostic pathway as all reviewers associated it with a diagnosis of essential thrombocythaemia; (b) increased amount of erythropoiesis and left shifting: when present, these were critical parameters as all reviewers associated them with a diagnosis of polycythaemia vera; (c) increased amount of granulopoiesis and left shifting: their occurrence are useful to rule out a diagnosis of essential thrombocythaemia; and (d) dense clusters of megakaryocytes: this parameter is clearly useful, because its presence is strongly suggestive for primary myelofibrosis. However, as dense clusters are infrequent, particularly in early-onset cases of PMF, it appears advisable to evaluate additional serial sections in order to increase the chance to detect them; (e) bulbous nuclei of megakaryocytes: when evident they suggest a diagnosis of primary myelofibrosis; (f) reticulin fibers: as expected, the presence of significant fibrosis frankly addresses the diagnosis towards primary myelofibrosis.

Although these figures do not allow us to design an independent histological diagnostic flow chart, it should be taken into account that some morphological parameters (such as cellularity, amount/shifting of erythropoiesis and granulopoiesis) could be more helpful than others in reaching the correct diagnosis.

The putative critical diagnostic role played by megakacaryocytes deserves to be slightly re-shaped. In fact, when the WHO classification 2001 was published, the concept that primary myelofibrosis was composed exclusively of bulbous megakaryocytes, while essential thrombocythaemia was represented almost by stag-horn ones was prevalent. Our daily experience actually suggests that megakaryocytes in these diseases are not so monomorphic and such a degree of morphological variability is always encountered, along with the presence of elements with additional morphological properties, including small elements, naked nuclei and maturation defects. The attempt should thus be to discover the typical ones in the whole biopsy setting, and this could be difficult particularly in early stage diseases.

Finally, to assess impact of subjectivity in the 'personal' diagnosis, we compared it with the 'consensus' diagnosis. Our results indicate an overall median percentage of crude agreement of 76% between the 'personal' diagnosis of each reviewers and 'consensus' diagnosis, with moderate-to-good values of Cohen's kappa statistic. These positive results certainly reflect the different experience among the four reviewers, but at the same time they confirm that it is possible to reach high levels of agreement and suggest that better results could probably be obtained by means of specific training. Taken together, these findings suggest that even if morphology alone allows the classification of most of the cases, integration with clinical and molecular data is mandatory for a correct classification of myeloproliferative neoplasms as indicated by the WHO classification.

In conclusion, the present study demonstrates that high levels of agreement could be reached among pathologists in recognizing the morphological features indicated by the WHO classification of Philadelphia chromosome-negative myeloproliferative neoplasms. Although this success reinforces the reproducibility of these histological parameters, about one-third of the cases of myeloproliferative neoplasms could not be efficiently classified by morphological examination alone. Notwithstanding these limitations, bone marrow histology should always be performed in the diagnostic work-up of myeloproliferative neoplasms, as individual parameters (in particular represented by overall cellularity, amount/shifting of erythropoiesis and granulopoiesis and bone marrow fibrosis) may be helpful in the differential diagnosis of Philadelphia chromosome-negative myeloproliferative neoplasms or have prognostic significance.[24,25] Keeping this in mind, and without entering into the logic of supporting or opposing the WHO classification, we believe that these promising results deserve to be further improved, possibly with the organization of larger consensus conferences, to ensure a more precise and early diagnosis for patients affected by these dramatic diseases.