SEARCH

SEARCH BY CITATION

Keywords:

  • myelofibrosis with myeloid metaplasia;
  • prognosis;
  • young patients

Abstract

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. References

Myelofibrosis with myeloid metaplasia (MMM) is an uncommon disorder in young individuals, for whom haemopoietic stem cell transplantation offers the only possibility of cure. However, although the latter procedure is associated with significant morbidity and mortality, the clinical course of MMM is variable, with some patients surviving for less than a year and others showing an indolent course. Selection of young MMM patients for transplantation or other newer therapies is currently difficult since no prognostic data exists for this subgroup. In the present collaborative study a number of initial clinical and laboratory parameters have been evaluated for prognosis in 121 MMM patients aged 55 years or less. Median survival of the series was 128 months (95% CI 90–172). In the Cox proportional hazard regression model three initial variables were independently associated with shorter survival: Hb <10 g/dl (P < 0.0001), the presence of constitutional symptoms (fever, sweats, weight loss) (P = 0.001), and circulating blasts ≥ 1% (P = 0.003). Based on the above three criteria, of the 116 patients with complete data, two groups were identified: a ‘low-risk’ group, characterized by 88 patients with up to one adverse prognostic factor, in whom MMM had an indolent course (median survival 176 months, 95% CI 130–188), and a ‘high-risk’ group, including 28 patients with two or three factors, who had a more aggressive disease (median survival 33 months, 95% CI 20–42). The above prognostic scoring system showed a high positive predictive value, sensitivity and specificity to predict survival in the series, and could be of help in making treatment decisions in young patients with MMM.

Myelofibrosis with myeloid metaplasia (MMM) is an infrequent chronic myeloproliferative disorder characterized by bone marrow fibrosis and extramedullary haemopoiesis ( Bouroncle & Doan, 1962; Burkhardt et al, 1986 ), in which the fibrosis is a reactive phenomenon secondary to a clonal haemopoietic disorder ( Jacobson et al, 1978 ). The disease affects mainly old people, with the median age at presentation of approximately 65 years in most series ( Dupriez et al, 1996 ; Cervantes et al, 1997 ; Reilly et al, 1997 ). Although patients with MMM survive for a median of 3.5–5 years ( Dupriez et al, 1996 ; Cervantes et al, 1997 ; Reilly et al, 1997 ; Barosi et al, 1988 ; Visani et al, 1990 ), there is a wide variation, with some patients surviving for decades. However, due to the low incidence of MMM in young individuals, information on the disease characteristics and prognosis in such a subpopulation is scarce ( Morel et al, 1997 ). Since no effective treatment for MMM is available, haemopoietic cell transplantation offers the only possibility of cure for young patients ( Guardiola et al, 1997 ; Anderson et al, 1997 ), but it is associated with significant morbidity and mortality. In addition, for young patients who lack a suitable donor the use of newer therapies may be justified ( Reilly, 1997). As a result, a better understanding of the disease characteristics and prognosis of younger MMM patients is important, not only to select those patients who may benefit from transplantation, but also to define the appropriate timing of the procedure, as well as to evaluate the possible survival benefit derived from newer therapeutic modalities.

To gain a better understanding of MMM and its prognosis in young patients, the main clinical and haematological features at diagnosis, as well as the clinical outcome, were analysed in 123 patients aged 55 years or younger presenting at four European centres.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. References

Patients and diagnostic criteria

Of the 550 patients consecutively diagnosed with MMM in four European hospitals, 123 were identified as 55 years old or younger at diagnosis. This subgroup represented 22.3% of the overall MMM population, and included 48 patients from the IRCCS Policlinico San Matteo, Laboratory of Medical Informatics (Pavia, Italy), 32 patients from the Service des Maladies du Sang, C.H.R.U. (Lille, France), 30 patients from the Hospital Clínic (Barcelona, Spain), and 13 from The Royal Hallamshire Hospital (Sheffield, U.K.). Final evaluation was restricted to 121 patients, after two cases with no follow-up were excluded. The diagnosis of MMM was based on the presence of an increased reticulin and/or collagen bone marrow content without any apparent cause, with the majority of cases exhibiting splenomegaly and a leucoerythroblastic picture. Patients with concurrent carcinoma, granulomatous disorders, metabolic osteopathy, lymphoma or myelodysplastic syndromes (especially proliferative forms of chronic myelomonocytic leukaemia) were excluded, as well as those with previous polycythaemia vera or other chronic myeloproliferative disorders, and acute myelofibrosis.

Treatment

23 patients received either no treatment or were managed with supportive therapy alone. The remaining 98 patients were given at least one of the following therapies: single-agent chemotherapy (hydroxyurea, 46 patients; busulphan, 22 cases; 6-mercaptopurine, six patients; cytarabine, four cases; pipobroman, two cases; and cyclophosphamide, melphalan, chlorambucil, and rubidazone, one case each), androgens (35 patients), prednisone (30 patients), alpha-interferon (eight patients), 1,25-dihydroxy-vitamin D (six cases), and erythropoietin (one case). 34 patients underwent splenectomy, usually because of unresponsive massive and symptomatic splenomegaly, and five additional patients received splenic irradiation. One patient underwent an allogeneic bone marrow transplantation 10 years from MMM diagnosis.

Bone marrow biopsies

A bone marrow trephine biopsy was performed in all cases as part of the initial evaluation, and subsequently stained with standard haematoxylin–eosin, silver stain for reticulin fibres, and trichrome for collagen and new bone formation assessment. Histological classification of MMM was as previously described ( Lennert et al, 1975 ).

Clinical and laboratory data evaluated

To avoid possible bias in the selection of prognostic variables, the following criteria applied: (a) they made clinical sense or had shown an association with survival in previous studies, and (b) they were available in the vast majority of patients. In all patients the following characteristics recorded at diagnosis were analysed for prognostic value: age (< or ≥ 46 years), gender, presence of constitutional symptoms (unexplained fever or excessive sweats persisting for more than 1 month, weight loss >10% of the baseline value in the year preceeding MMM diagnosis), spleen size (with the cut-off points being established at 5, 8 and 10 cm below the costal margin), hepatomegaly, Hb concentration (< or ≥ 10 g/dl), platelet count (< or ≥ 100 × 109/l), WBC count (< 4 × 109/l or >30 × 109/l), presence of circulating blasts (≥ 1%), circulating immature granulocytes (excluding blasts) ≤ or >10%, and bone marrow histological pattern. An evaluable bone marrow or unstimulated peripheral blood karyotype was obtained in 66 patients.

Statistical analysis

Diagnostic clinicohaematological variables (see above) were included in the Cox proportional hazard model ( Cox, 1972). Cytogenetic findings were excluded from analysis as data was available in only 50% of the patients. Those covariates that did not attain a significant association with survival were excluded from the model, and a new proportional hazard model was built with those remaining, with the process being repeated until all covariates in the model had a significant correlation with survival, as estimated by Wald's statistics ( Kleinbaum, 1996). Actuarial survival curves were traced according to the product-limit estimate of Kaplan & Meier (1958), and statistically compared by the Mantel-Cox test ( Mantel, 1966). The BMDP software was used for computations.

The usefulness of a prognostic score, derived from the above analysis, to predict an actual survival shorter than a selected cut-off time was assessed by calculating its positive predictive value, sensitivity and specificity, as previously described ( Cervantes et al, 1994 ). Positive predictive value denotes the accuracy of the prognostic score to predict a short survival (i.e. the probability that a patient who had been classified into the ‘high-risk’ group will actually have a short survival), whereas sensitivity represents the potency of the score to detect the short survivors (i.e. the proportion of patients classified as ‘high-risk’ among those with short survival). Specificity measures the proportion of patients classified as ‘low-risk’ among those who were actually long survivors.

Finally, three recently proposed prognostic scores for MMM were tested for their validity in our patient population: (a) the schema by Visani et al (1990 ), in which three prognostic groups were defined based on Hb value (< or ≥ 10 g/dl) and immature blood granulocytes < or ≥ 10%; (b) the Lille score ( Dupriez et al, 1996 ), which recognizes three risk groups based on two adverse prognostic factors (Hb < 10 g/dl, and WBC < 4 or >30 × 109/l); and (c) Cervantes et al (1997 ) score, which distinguishes two prognostic groups (low-risk, no or one adverse factor; and high-risk, two or three adverse factors), depending on the presence of constitutional symptoms, Hb < 10 g/dl, and circulating blasts ≥ 1%.

RESULTS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. References

Survival, acute transformation and causes of death

At the time of analysis (November 1997), 60/121 patients had died. Fig 1 shows the actuarial survival curve of the overall series, whose median survival was 128 months (95% CI 90–172). Acute transformation was observed in 16 patients, with the actuarial probability of evolution to acute leukaemia being 8.2% (95% CI 5.4–11) at 5 years from diagnosis and 19.8% (95% CI 14.8–24.8) at 10 years. In addition to those that transformed, the cause of death could be determined in another 36 patients, and included infection (17 cases), heart failure (six cases), bleeding (five cases), stroke (two cases), and pulmonary thromboembolism, hepatic failure due to post-splenectomy massive myeloid metaplasia of the liver, respiratory failure, cachexia, generalized oedema, and post-transplantation graft-versus-host disease (one case each).

image

Figure 1. MMM patients ≤ 55 years old.

Download figure to PowerPoint

Prognostic factors and risk-groups

The Cox proportional hazard regression identified five variables as independently associated with survival: Hb < 10 g/dl, presence of constitutional symptoms, circulating blasts ≥1%, circulating immature granulocytes >10%, and WBC < 4 or > 30 × 109/l. When the above selected variables were included in a new regression, the following three retained their significant association with survival: Hb < 10 g/dl, constitutional symptoms, and circulating blasts ≥1% ( Figs 2 A, 2B and 2C, and 2 Table II). Interestingly, in the subgroup of 66 patients with available cytogenetic studies no significant difference was observed when the survival of the 11 patients with chromosome abnormalities was compared with that of patients showing a normal karyotype.

image

Figure 2. Fig 2. Adjusted survival curves according to initial Hb value (A), constitutional symptoms (B) and circulating blast cells (C) at diagnosis. The dotted lines represent the base-line survival curves, i.e. the expected survival of the whole series if all patients had the same mean value for each explanatory covariate. The adjusted survival curves (continuous lines) show the expected change in the base-line survival when the corresponding covariate is settled at their maximal (or minimal) value whereas the other covariates remain unchanged.

Download figure to PowerPoint

Table 2. Table II. Initial covariates significantly associated with survival time at the multivariate analysis.Thumbnail image of

Based on the above selected prognostic variables, among the 116 patients with the data, four prognostic groups were identified, corresponding to patients with no, one, two or the three adverse prognostic factors, whose survival was significantly different (P < 0.0001, Fig 3A). However, overlapping was observed between the median survival 95% CI of patients with one or no adverse prognostic factors as well as between patients with two and three adverse prognostic factors ( Table III). In contrast, when the survival of patients with no or one poor prognostic factor was compared with that of patients with two or the three prognostic factors, a significant difference (P < 0.0001) was observed, but with no overlapping between the standard errors of the survival curves of both groups (Fig 3B). Because of this, two prognostic groups were finally chosen: a ‘low-risk’ group, characterized by 88 patients with no or one poor prognostic factors, whose median survival was 176 months (95% CI 153–199), and a ‘high-risk’ group, including 28 patients with two or three poor prognostic factors, with a median survival of 33 months (95% CI 20–42) (Fig 3B). These two prognostic groups were similarly distributed in the four series contributing to the study, with the percentage of ‘high-risk’ patients being 21% (Pavia), 27% (Lille), 26% (Barcelona) and 22% (Sheffield).

image

Figure 3. , n = 10). (B) Actuarial survival curves of ‘low-risk’ (n = 88) and ‘high-risk’ (n = 28) young MMM patients. The dotted lines indicate the standard error of the probability of survival.

Download figure to PowerPoint

Table 3. Table III. Survival of young MMM patients according to the number of adverse prognostic factors.Thumbnail image of

As can be seen in Fig 4, the above proposed prognostic scoring system showed a high positive predictive value, sensitivity and specificity to predict survival. Thus, at 5 years from diagnosis 72% of the patients in the ‘high-risk’ group had died whereas, conversely, 67% of the patients who had died at 5 years belonged to the ‘high-risk’ group, and 89% of those surviving for > 5 years had been classified within the ‘low-risk’ group.

image

Figure 4. Fig 4. Positive predictive value, sensitivity and specificity of the prognostic scoring system to predict the survival of the patients at different cut-off times from diagnosis.

Download figure to PowerPoint

Since many groups use a lower age limit for transplantation with matched unrelated donors (≤ 45 years in most centres), we therefore assessed the validity of the prognostic scoring system in the 48 patients in the series who were 45 years old or younger, of whom 39 patients were classified in the ‘low-risk’ group and nine in the ‘high-risk’ group. Median survival of these two subgroups of patients was 145 (95% CI 121–212) and 22 months (95% CI 14–30), respectively, with the difference being statistically significant (P < 0.0001).

Application of other prognostic scores

Both Visani's ( Visani et al, 1990 ) and the Lille score ( Dupriez et al, 1996 ) identified a low-risk group of patients with similar median survival: 188 months (95% CI 145–216) (n = 63 patients), and 176 months (95% CI 128–212) (n= 68), respectively. However, they were not so successful in separating the intermediate- and high-risk groups ( Figs 5A and 5B). When patients from the latter two risk-groups were combined, the discriminating power of the above scores improved, allowing the identification of a new risk-group, including 53 and 51 patients, respectively, whose median survival was 66 months in both cases (95% CI 37–114 months for Visani's score and 35–119 months for the Lille score). Finally, as expected (since it is based on the same prognostic factors emerging from the present analysis), Cervantes' prognostic score ( Cervantes et al, 1997 ) clearly identified a low- and a high-risk groups of younger MMM patients (see Fig 3B).

image

Figure 5. Fig 5. (A) Actuarial survival curves of young MMM patients according to the Visani et al (1997) prognostic scoring system. (B) Actuarial survival curves of the patients according to the Lille prognostic score ( Dupriez et al, 1996 ).

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. References

Myelofibrosis with myeloid metaplasia (MMM), also known as idiopathic or primary myelofibrosis, is a chronic myeloproliferative disorder characterized by the presence of increased bone marrow fibrosis as a secondary phenomenon to the neoplastic proliferation of a pluripotent haemopoietic stem cell ( Jacobson et al, 1978 ). The disease is heterogenous in its presentation and clinical course, with median survival ranging from 3.5 to 5.5 years in recent series, although there is a wide individual variation ( Dupriez et al, 1996 ; Cervantes et al, 1997 ; Reilly et al, 1997 ; Barosi et al, 1988 ; Visani et al, 1990 ). Although MMM is considered a disease of the elderly, a number of young patients have been reported in all series. In recent years various studies have identified a number of prognostic factors in MMM, and, as a result, several prognostic scoring systems have been proposed ( Dupriez et al, 1996 ; Cervantes et al, 1997 ; Reilly et al, 1997 ; Barosi et al, 1988 ; Visani et al, 1990 ). However, there is little information on young MMM patients ( Barosi et al, 1988 ; Morel et al, 1997 ), although the identification of prognostic factors and risk-groups in this subgroup is of great clinical importance. This is because, whereas no effective therapy for MMM is currently available ( Reilly, 1997), allogeneic bone marrow transplantation offers a potential cure for young patients ( Guardiola et al, 1997 ; Anderson et al, 1997 ). However, both the substantial morbidity and the 30% mortality associated with the procedure, and the indolent clinical course of many MMM patients, makes the choice and timing of transplantation difficult ( Guardiola et al, 1997 ; Anderson et al, 1997 ). As a result, a better understanding of the prognosis in young patients with MMM is needed to enable treatment strategies to be designed for these patients. In addition, the identification of prognostic factors may be of help in evaluating the survival benefit that could result from the use of experimental treatment modalities ( Reilly, 1997).

In view of the above, we have undertaken a prognostic study of young MMM patients aged 55 years or younger, the age limit currently used by most transplant centres. Using this definition, 123 patients were identified from a total of 550 cases, representing almost a quarter of the general MMM population. The initial clinico-haematological characteristics of the 121 evaluable patients fitted the typical picture of MMM, except for a lower frequency of chromosome abnormalities when compared to that reported for the general MMM population ( Dupriez et al, 1996 ; Cervantes et al, 1997 ; Reilly et al, 1997 ; Demory et al, 1988 ). The most striking feature of young MMM patients was, however, their prolonged survival, with median survival being more than twice that reported in recent MMM series ( Dupriez et al, 1996 ; Cervantes et al, 1997 ; Reilly et al, 1997 ; Barosi et al, 1988 ; Visani et al, 1990 ). This longer survival of younger patients is also seen in other haematological neoplasias ( Montserrat et al, 1991 ), and probably reflects a better host's physical state. However, no differences were observed in either the frequency of acute transformation or the causes of death in this patient subgroup ( Dupriez et al, 1996 ; Cervantes et al, 1997 ).

In the current study three presenting features were associated with survival in young patients with MMM (anaemia, presence of constitutional symptoms, and circulating blasts). The prognostic importance of anaemia in MMM has been universally recognized ( Dupriez et al, 1996 ; Cervantes et al, 1997 ; Reilly et al, 1997 ; Barosi et al, 1988 ; Visani et al, 1990 ; Varki et al, 1983 ; Njoku et al, 1983 ; Manoharan, 1988; Rupoli et al, 1994 ). With regard to constitutional symptoms, their unfavourable prognostic influence has been demonstrated in all studies in which such clinical features were evaluated ( Dupriez et al, 1996 ; Cervantes et al, 1997 ; Varki et al, 1983 ; Njoku et al, 1983 ), and the prognostic value of circulating blasts has also been emphasized ( Dupriez et al, 1996 ; Cervantes et al, 1997 ). The lack of prognostic value of bone marrow histological findings is in agreement with the results from previous studies, both by us and others ( Varki et al, 1983 ; Pereira et al, 1990 ). In contrast, the prognostic significance of chromosomal abnormalities ( Reilly et al, 1997 ; Demory et al, 1988 ; Lawler & Swansbury, 1985) could not be confirmed, although this may reflect the small number of patients with this finding.

In the past, several attempts at defining prognostic groups in MMM have been carried out ( Dupriez et al, 1996 ; Cervantes et al, 1997 ; Reilly et al, 1997 ; Barosi et al, 1988 ; Visani et al, 1990 ; Njoku et al, 1983 ; Rupoli et al, 1994 ). Among them, those by Visani et al (1990 ) and the Lille score ( Dupriez et al, 1996 ), based on the analysis of a large number of patients, are noteworthy. Visani's scoring system is based on the presence of two adverse prognostic factors (Hb < 10 g/dl and blood granulocyte precursors >10%), and separates a low-, an intermediate- and a high-risk groups in MMM, with median survival of 81, 39 and 31 months, respectively. The Lille score, based on two features (Hb <10 g/dl and WBC < 4 or >30 × 109/l), similarly identifies three prognostic groups, with median survivals of 93, 26 and 13 months. When applied to young patients, the above scoring systems clearly identified a low-risk group of patients, but were unsuccessful in separating the intermediate- and high-risk groups. The discriminating power of both scores in the present series, however, improved when the intermediate- and high-risk groups were pooled to generate a new high-risk group of patients. The prognostic staging system recently proposed by Reilly et al (1997 ) combines age, Hb value and karyotype to identify good- and poor-risk groups in MMM. However, the low frequency of chromosome alterations observed in young MMM patients appears to limit the value of this approach. Finally, the prognostic scoring system by Cervantes et al (1997 ) enables a clearcut separation of low- and high-risk IM patients, on the basis of the presence or absence of constitutional symptoms, Hb < 10 g/dl and circulating blasts.

In the present collaborative study, in which a large number of young MMM patients were analysed, prognostic factors were identified which allowed the separation of ‘low-risk’ and ‘high-risk’ groups. The former included three-quarters of the patients, with a median survival of > 14 years, whereas the latter contained a quarter of the patients, with a median survival of < 3 years. No overlapping was observed between the standard errors of the two survival curves. The high positive predictive value, sensitivity and specificity of this staging system makes it of value for the identification of patients suitable for haemopoietic stem cell transplantation or other newer therapeutical strategies. Moreover, the above prognostic system proved to be of value also in the subgroup of younger patients (≤ 45 years), who are potential candidates for transplantation from unrelated donors.

In conclusion, the present study shows that the combination of three simple clinical and haematological features at diagnosis can provide a prognostic model in young patients with MMM. Such a schema could facilitate the therapeutic decision in such patients.

Acknowledgements

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. References

This work has been supported in part by the grants SGR 1995/052 from the Generalitat de Catalunya, FISS 96/0274 from the Spanish Ministry of Health, and Maderas de Llodio S.A.L.

References

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. References
  • 1
    Anderson, J.E., Sale, G., Appelbaum, F.R., Chauncey, T.R, Storb, R. (1997) Allogeneic marrow transplantation for primary myelofibrosis and myelofibrosis secondary to polycythaemia vera or essential thrombocytosis. British Journal of Haematology, 98, 1010 1016.
  • 2
    Barosi, G., Berzuini, C., Liberato, L.N., Costa, A., Polino, G, Ascari, E. (1988) A prognostic classification of myelofibrosis with myeloid metaplasia. British Journal of Haematology, 70, 397 401.
  • 3
    Bouroncle, B & Doan, C.A. (1962) Myelofibrosis: clinical, haematologic and pathologic study of 110 patients. American Journal of Medical Sciences, 243, 697 715.
  • 4
    Burkhardt, R., Bartl, R., Frisch, B., Ketter, G., Mahl, G, Sund, M. (1986) Working formulation of chronic myeloproliferative disorders based on histological, haematological, and clinical findings. Journal of Clinical Pathology, 39, 237 252.
  • 5
    Cervantes, F., Pereira, A., Esteve, J., Rafel, M., Cobo, F., Rozman, C, Montserrat, E. (1997) Identification of ‘short-lived’ and ‘long-lived’ patients at presentation of idiopathic myelofibrosis. British Journal of Haematology, 97, 635 640.
  • 6
    Cervantes, F., Robertson, J.E., Rozman, C., Baccarani, M., Tura, S., Gómez, G.A., Braun, T.J., Clarkson, B.D., Pereira, A. (1994) Long-term survivors in chronic granulocytic leukaemia: a study by the International CGL Prognosis Study Group. British Journal of Haematology, 87, 293 300.
  • 7
    Cox, D.R. (1972) Regression models and life tables. Journal of the Royal Statistical Society (B), 34, 187 220.
  • 8
    Demory, J.L., Dupriez, B., Fenaux, P., Lai, J.L., Beuscart, R., Jouet, J.P., Deminatti, M, Bauters, F. (1988) Cytogenetic studies and their prognostic significance in agnogenic myeloid metaplasia: a report on 47 cases. Blood, 72, 855 959.
  • 9
    Dupriez, B., Morel, P., Demory, J.L., Lai, J.L., Simon, M., Plantier, I, Bauters, F. (1996) Prognostic factors in agnogenic myeloid metaplasia: a report on 195 cases with a new scoring system. Blood, 88, 1013 1018.
  • 10
    Guardiola, Ph., Esperou, H., Cazals-Hatem, D., Ifrah, N., Jourt, J.P., Buzyn, A., Sutton, L., Gratecos, N., Tilly, H., Lioure, B, Gluckman, E. (1997) Allogeneic bone marrow transplantation for agnogenic myeloid metaplasia. British Journal of Haematology, 98, 1004 1009.
  • 11
    Jacobson, R.J., Salo, A, Fialkow, P.J. (1978) Agnogenic myeloid metaplasia: a clonal proliferation of hemopoietic stem cells with secondary myelofibrosis. Blood, 5, 189 194.
  • 12
    Kaplan, E.L & Meier, P. (1958) Nonparametric estimations from incomplete observations. Journal of the American Statistical Association, 53, 457 481.
  • 13
    Kleinbaum, D.G. (1996) Survival Analysis, pp. 83 128. Springer, New York.
  • 14
    Lawler, S.D & Swansbury, G.J. (1985) Cytogenetic studies in myelofibrosis and related conditions. Myelofibrosis: Pathophysiology and Clinical Management (ed. by S. M. Lewis), pp. 167 194. Marcel Dekker, New York.
  • 15
    Lennert, K., Naghai, K, Schwarze, E.W. (1975) Patho-anatomical features of the bone marrow. Clinics in Haematology, 4, 331 351.
  • 16
    Manoharan, A. (1988) Myelofibrosis: prognostic factors and treatment. British Journal of Haematology, 69, 295 298.
  • 17
    Mantel, N. (1966) Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Chemotherapy Reports, 50, 163 170.
  • 18
    Montserrat, E., Gomis, F., Vallespí, T., Ríos, A., Soler, J., Alcalá, A., Morey, M., Ferrán, C., Díaz-Mediavilla, J., Flores, A., Woessner, S., Batlle, J., González-Aza, C., Rovira, M., Reverter, J.C, Rozman, C. (1991) Presenting features and prognosis of chronic lymphocytic leukemia in younger adults. Blood, 78, 1545 1551.
  • 19
    Morel, P., Demory, J.L, Dupriez, B. (1997) Relevance of prognostic features in myeloid metaplasia to selection of patients for bone marrow transplantation. (Letter). Blood, 89, 2219 2220.
  • 20
    Njoku, O.S., Lewis, S.M., Catovsky, D, Gordon-Smith, E.C. (1983) Anaemia in myelofibrosis: its value in prognosis. British Journal of Haematology, 54, 79 89.
  • 21
    Pereira, A., Cervantes, F., Brugués, R, Rozman, C. (1990) Bone marrow histopathology in primary myelofibrosis: its value in prognosis. European Journal of Haematology, 44, 94 98.
  • 22
    Reilly, J. (1997) Myelofibrosis: pathogenesis, natural history and management. Blood Reviews, 11, 233 242.
  • 23
    Reilly, J.T., Snowden, J.A., Spearing, R.L., Fitzgerald, P.M., Jones, N., Watmore, A, Potter, A. (1997) Cytogenetic abnormalities and their prognostic significance in idiopathic myelofibrosis: a study of 106 cases. British Journal of Haematology, 98, 96 102.
  • 24
    Rupoli, S., Da Lio, L., Sisti, S., Campanati, G., Salvi, A., D'amico, S., Cincipirini, A, Leoni, P. (1994) Primary myelofibrosis: a detailed statistical analysis of the clinicopathological variables influencing survival. Annals of Hematology, 68, 205 212.
  • 25
    Varki, A., Lottenberg, R., Griffith, R, Reinhard, E. (1983) The syndrome of idiopathic myelofibrosis: a clinico-pathologic review with emphasis on the prognostic variables predicting survival. Medicine (Baltimore), 62, 353 371.
  • 26
    Visani, G., Finelli, C., Castelli, U., Petti, M.C., Ricci, C., Vianelli, L., Gianni, L., Zuffa, E., Aloe Spiriti, M.A., Latagliata, R., Pileri, S., Magrini, U., Gugliota, L., Morra, E., Bernasconi, C., Mandelli, F, Tura, S. (1990) Myelofibrosis with myeloid metaplasia: clinical and haematological parameters predicting survival in a series of 133 patients. British Journal of Haematology, 75, 4 9.