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

Disclosures

Mod Pathol. 2014;27(6):814-822. 

In This Article

Materials and Methods

A series of 113 patients from four Institutions (Milano, Bologna, Reggio Emilia, Palermo) referring to the Italian Registry of Thrombocythaemias (Registro Italiano Trombocitemie, RIT) were collected for the study. Consecutive patients, with newly diagnosed Philadelphia chromosome-negative myeloproliferative neoplasms, fulfilling the WHO classification criteria and diagnosed in 2004 were included in this study; this choice was made in order to have a reliable clinical follow-up. All of the patients signed their informed consent. As our aim focused on the differential diagnosis among Philadelphia chromosome-negative myeloproliferative neoplasms, cases with)[11] on the basis of the local diagnosis were excluded from the study. For each patient, bone marrow biopsy specimen at diagnosis was available and stained with hematoxylin–eosin, Giemsa and Gomori's silver impregnation.

Two groups of pathologists reviewed all the cases. The first group composed by eight pathologists (UG, CT, MP, AMF, GFO, SA, EBon, VF) centrally reviewed the cases and reached a 'consensus' diagnosis at a multi-headed microscope based on the morphological analysis and the complete knowledge of the clinical data. Only cases for which a 100% agreement was reached were included in this study. A total of 103 out of 113 cases were selected and classified as follows: essential thrombocythaemia: n=34, primary myelofibrosis: n=44 (myelofibrosis grade 0: n=15, myelofibrosis grade 1: n=16, myelofibrosis grade 2: n=13) and polycythaemia vera: n=25 (all in the polycythaemic phase of the disease). Ten cases were excluded from the study because of the inadequacy of the material (n=6) or because they have been classified as myeloproliferative neoplasm 'unclassifiable' (n=2) and as pre-polycythaemic phase of polycythaemia vera (n=2). Table 1 summarizes the most important clinical data of the study population.

The second group of four pathologists (ES, EBov, AM, RV) were asked to review individually all the 103 selected cases, in a blinded fashion, without any knowledge of the 'consensus' diagnosis made by the first group and of the clinical data, with the exception of the patient's age. Their aim was to recognize and register the main morphological parameters included in the WHO classification for each case and report their results in a database. To carry out this task, each reviewer received by mail the same box containing the original slides evaluated by the first panel of pathologists and a histological form reporting the list of the morphological features with minimal explanation on how to compile it, as reported in the following paragraphs.

The following 18 morphological variables were examined: (1) overall bone marrow cellularity in relation to patient's age; (2) amount of erythropoiesis; (3) left-shifting erythropoiesis; (4) amount of granulopoiesis; (5) left-shifting granulopoiesis; (6) myeloid to erythroid ratio; (7) amount of megakaryocytes; (8) loose clusters of megakaryocytes, defined as aggregation of ≥3 megakaryocytes, albeit not in reciprocal contact; (9) dense clusters of megakaryocytes, defined as aggregation of ≥3 megakaryocytes in reciprocal contact; (10) pleomorphism of the clusters of megakaryocytes, defined as the presence of a cluster of ≥3 megakaryocytes of variable size, ranging from small to giant; (11) small megakaryocytes; (12) giant megakaryocytes (13) megakaryocyte nuclear hyperlobulation ('stag horn-like' morphology); (14) megakaryocyte bulbous nuclei ('cloudy-like' or 'balloon-shaped' morphology); (15) megakaryocyte naked nuclei; (16) megakaryocyte maturation defect defined as nuclear-to-cytoplasmic ratio alteration; (17) megakaryocyte nuclear dysmorphism (hyperchromatic and dysmorphic nuclei); and (18) entity of marrow fibrosis, (determined according to the European consensus grading system for marrow fibrosis).[11] The four[12] To register a qualitative variable as 'present', the reviewer had to identify that specific morphological features at least two times in the same slide.

Figure 1 represents some of these morphological features.

Figure 1.

(a) A case of polycythaemia vera with prominent left-shifting erythropoiesis (Giemsa, × 40). (b) Dense cluster of megakaryocytes sometimes with bulbous nuclei are evident in this case of primary myelofibrosis (Giemsa, × 20). (c) Large-to-giant megakaryocytes with hyperlobulated nuclei in a case of essential thrombocythaemia (H&E, × 20). (d) A pleomorphic cluster of megakaryocytes is evident in a case of polycythaemia vera (H&E, × 20).

In order to assess how morphology alone could help to reach the correct diagnosis, the four reviewers were subsequently asked to propose a 'personal' diagnosis by just analyzing and assembling the data collected in their databases, without reviewing the case under the microscope.

Statistical Analysis

To investigate the relationship between morphological profile of each case, 'personal' and 'consensus' diagnosis, we performed a multiple correspondence analysis. This is a data analysis technique for categorical data, used to detect and represent underlying structures in a data set. It achieves the results by representing data as points in a low-dimensional Euclidean space. In this study, we only considered the first two axes. As a final result of multiple correspondence analysis, a two-dimensional map was produced where diagnoses, reviewers and category of each morphological feature were represented as points. The degree of accuracy of the map is given by the percentage of total variance represented on the two axes of the map. In Figure 2, the percentage of variance explained by the two axes is about 91% (horizontal axis: 63.5%; vertical axis: 27.7%).

Figure 2.

Two dimension graphical representation of Multiple Correspondence Analysis. The horizontal axis contrasts the morphological profiles of essential thrombocythaemia (ET) patients vs those with primary myelofibrosis (PMF), while the vertical axis contrasts polycythaemia vera (PV) vs ET and PMF. Quantitative variables (♦=reduced, ▴=normal, ▪=increased) and qualitative variables (▵=absent, ▴=present) are represented.

Cohen's kappa statistic was used to express the agreement between the 'personal' diagnosis made by each reviewer and 'consensus' diagnosis.

Asymptotic 95% confidence intervals for kappa statistic were computed according to Fleiss et al.[13]

A kappa of 0 means that the agreement is no better than that expected by chance alone, and kappa values of 1.00 indicate perfect agreement. By convention, a kappa between 0.81 and 1.00 is interpreted as indicating excellent agreement. Values of <0.20, 0.21–0.40, 0.41–0.60 and 0.61–0.80 are interpreted as showing poor, fair, moderate and good agreement, respectively.[14] All data processing and analyses were carried out with the SAS statistical software (version 9.2; SAS Institute, Cary, NC, USA).

processing....