Trend of incidence, subsite distribution and staging of colorectal neoplasms in the 15-year experience of a specialised cancer registry.

BACKGROUND: Two-thirds of colorectal malignancies are localised in the left colon and rectum. Recent studies suggest a trend towards an increase of right-sided tumours which might have important implications for screening and surveillance. A colorectal cancer registry was set up in Modena, northern Italy, with the purpose of examining incidence, subsite distribution and staging of colorectal malignancies over a 15-year period. PATIENTS AND METHODS: From 1984 to 1998, 2517 tumours in 2462 patients were detected and staged with the tumour node metastasis (TNM) system. The 'right colon' was considered from caecum to splenic flexure; the 'left colon' included descending and sigmoid colon; and the 'rectum' included rectosigmoid junction, ampulla and anus. RESULTS: Cancer incidence showed an overall increase. Considering the various subsites, an increase of 33.7% in all colonic segments was shown whereas rectal tumours tended to decline. TNM staging showed a gradual increase of localised lesions (41.2% in 1984 versus 53.3% in 1998), with a proportional reduction of advanced tumours. CONCLUSIONS: Our study indicates an increase of tumour incidence in all colonic segments more than a shift to the right colon. TNM staging tended to improve with an appreciable increase of localised lesions. These findings could be consequent to a more extensive use of colonoscopy.

Response to first-line chemotherapy and long-term survival in patients with multiple myeloma: results of the MM87 prospective randomised protocol.

In this study we evaluated whether a good response to conventional chemotherapy, i.e. a significant tumour reduction, is a prerequisite for improved survival in multiple myeloma (MM). Between January 1987 and March 1990, 341 consecutive previously untreated patients with MM received chemotherapy within the prospective, multicentre, randomised Protocol MM87. Of these, 258 patients were evaluable for both response and long-term survival and 244 (94.6%) have died. The median survival of all patients was 40 months (6-162 months). The median survival did not differ between patients who had complete response (CR) (50 months (9-162 months)), partial response (PR) (46 months (8-147 months)) or stable disease (SD) (41 months (7-135 months)). The median survival was shorter (13.6 months (6-135 months)) (P<0.0001) in patients whose disease progressed while they were receiving first induction chemotherapy. Causes of death were more frequently (P=0.04) related to MM in patients who had progressive disease (PD) than in patients who had a CR or PR or SD. The main clinical and laboratory characteristics were similar in the four groups. These data indicate that patients who maintain SD during first-line chemotherapy have a prognosis similar to that of patients who attain a response. Only patients whose disease progresses have a distinctly worse outcome.

Expression of p53, bcl-2 and ras oncoproteins and apoptosis levels in acute leukaemias and myelodysplastic syndromes.

We analysed by immunocytochemistry the expression of p53, bcl-2 and ras proteins in bone marrow blasts from 59 patients with acute leukaemia (AL), 36 myeloid (AML) and 23 lymphoid (ALL), and from 22 patients with myelodysplastic syndrome (MDS); our aim was to examine if abnormalities in their expression were associated with peculiar biological and clinical findings, or with an altered apoptosis rate, as measured by TUNEL technique. The oncoproteins were expressed with extreme variability, without significant differences among the various morphological or immunological AL subtypes. The mean percentages of bcl-2+ blasts were significantly higher in AML than in MDS (p = 0.01), and in MDS with bone marrow blastosis than in the forms without excess of blasts (p = 0.007). The lowest percentages of apoptotic cells were observed in ALL (mean 1%, p = 0.006), whereas in MDS the apoptotic index was higher (16.7%) than in AML (8.6%) and than in the normal controls (10.8%). but the difference tended to be statistically significant only for cases of refractory anaemia. Whereas in AML and MDS the apoptotic rate was independent of the oncoprotein expression, in ALL there was a significant linear relationship between TUNEL and ras positivity (p = 0.01). Among AML patients treated with intensive polychemotherapy, no differences were observed in oncoprotein expression and apoptotic rate between responders and resistant cases. In conclusion, our data are in agreement with the hypothesis that decreased apoptosis and enhanced cell survival are associated with AL, whereas a high level of apoptosis may be responsible for the ineffective hematopoiesis in MDS; abnormal expression of oncoproteins, even if not strictly related to apoptosis level, may influence disease behaviour.

A modified high-dose dexamethasone regimen for primary systemic (AL) amyloidosis.

High-dose dexamethasone (HD-Dex) has been reported to benefit AL amyloidosis patients with varying response rates. Our preliminary experience with the usual HD-Dex schedule indicated that the induction phase was rather toxic in AL patients. We therefore adopted a milder schedule consisting of dexamethasone 40 mg on d 1-4 q21 d for up to eight cycles. Overall 8 out of 23 (35%) treated patients responded to treatment in a median time of 4 months (range 2-6 months) without significant toxicity. This regimen may be considered front-line therapy when autologous stem cell transplantation is not feasible and when a rapid response is particularly important.

Comparison of prognostic models in patients with advanced Hodgkin disease. Promising results from integration of the best three systems.

BACKGROUND: Several prognostic systems have been elaborated for patients with Hodgkin disease (HD) over the last 12 years, but early identification of a reasonably large group of both low and high risk, advanced stage patients remains unsatisfactory. METHODS: Seven well known models were applied to 516 patients with advanced HD, with 315 patients used for the study sample and 201 patients used for the test sample. Individual performances as well as joint performances were analyzed univariately and multivariately in relation to overall survival, recurrence free survival, and time to treatment failure by means of a proportional hazards model. RESULTS: None of the models identified a group containing > 10% of patients from the total population who had a failure risk of either < or = 10% or > or = 50%. The systems of the International Database on Hodgkin Disease, the Memorial Sloan-Kettering Cancer Center, and the International Prognostic Factor Project showed the best prognostic power; only these three, when analyzed together, predicted clinical outcome with a statistically significant fit to the clinical data. Integration of the three systems in a linear model dramatically improved their individual discriminatory capacity by identifying patients with 10% and 50% failure risks, respectively, in 23% and 24% of the study patient population and in 19% and 25% of the test population, respectively. CONCLUSIONS: As powerful and simple new prognostic factors are awaited that may improve our predictive ability, this integrated index is probably the best way to exploit the significance of those presently available. The program required for the calculations can be downloaded from the Internet at the web site http://www.unimo.it/gisl/default.htm.

Image-aided estimate of tumor burden in Hodgkin's disease: evidence of its primary prognostic importance.

PURPOSE: To explore a more direct method for evaluating tumor burden (TB) in Hodgkin's disease (HD) and to verify its prognostic importance. PATIENTS AND METHODS: The volume of TB at diagnosis was directly and retrospectively measured in 121 HD patients through images of the lesions recorded by computed tomographic (CT) scan of the thorax, abdomen, and pelvis for all deep sites of involvement and many superficial ones, and by ultrasonography (US) for the remaining superficial lesions. RESULTS: The TB, which was obtained from the sum of the volumes of all the lesions measured on CT scans and US and normalized to body-surface area (relative TB [rTB]), showed a median value of 102.6 cm(3)/m(2) (range, 2.2 to 582.8). At multivariate analysis for prognostic value, rTB was the parameter that statistically correlated best with time to treatment failure (P = 2.2 x 10(-6)), followed by erythrocyte sedimentation rate (ESR) (P =.0003), and serum fibrinogen (P =.0112). The prognostic discrimination allowed by rTB alone proved to be clearly superior to that obtained with the score of the International Prognostic Factor Project. The rTB was found to be correlated with many clinical staging parameters (bulky disease, number of involved lymph node regions, serum lactate dehydrogenase, ESR, hemoglobin, Karnofsky index), but its predictability from these variables was low (R(2) =.668). CONCLUSION: Relative TB is emerging as a strong prognostic factor in HD, more powerful than and largely independent of those hitherto known and used. Further studies are needed to confirm these results and exploit their clinical value, particularly the relationship among rTB, drug doses, and response.

Biological features of the clone involved in primary amyloidosis (AL).

Primary light chain-associated amyloidosis (AL) is a plasma cell dyscrasia that causes morbidity via systemic tissue deposition of monoclonal light chains in the form of fibrils (amyloid). It is the most common form of systemic amyloidosis in Western countries and is rapidly fatal. Knowledge of the pathobiology of the underlying B cell clone is of primary importance for the design and optimization of therapeutic strategies.

Lomustine and melphalan cannot be replaced by cyclophosphamide and etoposide without reducing efficacy in MOPPEBVCAD chemotherapy for advanced Hodgkin's disease.

BACKGROUND AND OBJECTIVES: To evaluate the feasibility, toxicity and preliminary results of a potentially less toxic variant of the MOPPEBVCAD chemotherapy regimen for advanced Hodgkin's disease: MOPPEBVCyED, in which cyclophosphamide and etoposide replace lomustine and melphalan, respectively, with the remaining components being unaltered. DESIGN AND METHODS: The study was multicenter, prospective and randomized, and enrolled 67 patients with newly diagnosed stage IIB, III, IV Hodgkin's disease (62 were expected on the grounds of statistical considerations). Radiotherapy was restricted to sites of bulky involvement or to areas that responded incompletely to chemotherapy. Median follow-up was 48 months. RESULTS: Comparing MOPPEBVCAD vs. MOPPEBVCyED, the results were as follows: complete remissions 35/35 vs. 30/32 (plus one partial remission and one disease progression); relapses 5 vs. 8; deaths 2 (one of myelodysplasia) vs. 2; delivered mean dose intensity (DI): lomustine 0.79+/-0.67 vs. cyclophosphamide 0.82+/-0.32; melphalan 0.80+/-0.13 vs. etoposide 0.86+/-0.18; average DI of the 7 drugs common to both regimens 0.73+/-0.10 vs. 0.83+/-0.11; all 9 drugs 0.75+/-0.13 vs. 0.84+/-0.09 (p=0.002); projected 5-year failure-free survival 0.79 vs 0.62; second cancers, two myelodysplasias vs. one carcinoma of the kidney. Toxicities were not statistically different except for heavier thrombocytopenia being recorded with MOPPEBVCAD. INTERPRETATION AND CONCLUSIONS: The higher cumulative and single drug DI recorded with MOPPEBVCyED may reflect better short-term tolerability, but it does not lead to better disease control. Its late toxicity may be expected to be lower in the future but at present it does not seem to be a sufficient reason to substitute MOPPEBVCyED for MOPPEBVCAD.

Cell cycle status and apoptosis of hematopoietic progenitor cells released into the peripheral blood after taxanes and granulocyte colony-stimulating factor in breast cancer patients.

Paclitaxel and its analogue docetaxel show a significant antitumor activity, particularly evident in breast cancer. Paclitaxel has also been proved to be effective as a peripheral blood progenitor cell (CPC) mobilizing agent. To optimize the use of active, disease-specific drugs as CPC priming, we have evaluated the effects of either paclitaxel or docetaxel both at standard dosages and followed by granulocyte colony-stimulating factor (G-CSF), on circulating CPC release and function in 18 patients with advanced breast cancer who had failed previous anthracycline-based regimens. The reported differences in biological behaviour between bone marrow and blood-derived hematopoietic progenitor cells and the ability of both paclitaxel and docetaxel to induce apoptosis, prompted us to simultaneously evaluate the cell cycle perturbations induced on CD34+ cells. Median CD34+ peaks were 24 microl (range: 10-58) in the paclitaxel-treated patients and 39 microl (range: 17-91), respectively, in the patients who received docetaxel. After paclitaxel, the percentage of CD34+ cells in S-phase was low (bromodeoxyuridine, BrdU, labelling index = 3.4+/-2%) with a concomitant presence of early apoptotic cells (8.1+/-3%). On the contrary, after docetaxel, the percentage of CD34+ cells in S-phase was higher (BrdU labelling index = 14.5+/-4%, p<0.05 vs. paclitaxel), while early apoptotic cells were detected at a similar rate (8. 6+/-3%, p = n.s. vs. paclitaxel). In conclusion, when used at standard dosages, with respect to paclitaxel + G-CSF, docetaxel + G-CSF is a more satisfactory tool to mobilize CPC and to induce them into the cell cycle. These data should be taken into account when combinations of docetaxel with other agents are explored as CPC mobilizing regimens for autografting.

The role of surgery in the treatment of gastrointestinal lymphomas other than low-grade MALT lymphomas.

BACKGROUND AND OBJECTIVE: A bias in clinical investigations on gastrointestinal lymphomas is the lack of testing the intention to treat as to resection, emergency conditions at presentation and selection brought about by the evaluation of feasibility of surgery. DESIGN AND METHODS: A prospective study involved 154 patients with gastrointestinal nodular or high-grade MALT lymphomas, 111 with a gastric and 43 with an intestinal presentation. The decision to resect or treat conservatively was left to clinicians, on condition that it was previously defined for each patient. RESULTS: Failure-free survival was significantly higher in the 106 resected patients than in the 48 unresected ones but did not differ according to either primary intention to treat or emergency surgery/elective treatment. Survival was similar in patients operated on by choice and in those because of an emergency. Intentionally unresected patients had a significantly better survival than those not undergoing surgery despite the initial intention, for a number of clinical reasons. Patients with gastric lymphoma survived longer than those with intestinal disease and prognostic factors were analyzed separately in the two groups. The best predictors of prognosis were performance status and serum lactic dehydrogenase level in gastric lymphomas, resection alone in intestinal ones. INTERPRETATION AND CONCLUSIONS: The prognosis of gastric lymphomas depends on lymphoma-related factors and not on surgical treatment. The prognosis of intestinal ones is exclusively related to surgery. These data support the appropriateness of different clinical approaches to gastric and intestinal lymphomas.

A variant of ProMACE-CytaBOM chemotherapy for non-Hodgkin's lymphoma with threefold higher drug dose size but identical cumulative dose intensity. A pilot study of the Italian lymphoma study group (GISL).

BACKGROUND AND OBJECTIVE: The positive results of high-dose chemotherapy followed by rescue with bone marrow progenitor cell transplantation are generally ascribed to the high dose size (DS) of the drugs given. However, a concomitant marked increase in dose intensity (DI) is always involved. With the aim of comparing the role of DS and DI in non-Hodgkin's lymphomas, a variant of Fisher's ProMACE-CytaBOM regimen was designed in which the projected cumulative drug DIs remained the same as in the original schedule but the DSs were tripled. DESIGN AND METHODS: Dosages in mg/m(2), route and days of administration were the following: cyclophosphamide 1,950 i.v. on days 1, 64; methotrexate 360 i.v. days 15, 78; vincristine 1.4 iv days 15, 78, 43, 106; etoposide 360 i.v. days 29, 92; epirubicin 120 i.v. days 29, 92; bleomycin 15 i.v. days 43, 106; cytarabine 900 i.v. days 50, 113. Thirty-six outpatients with intermediate- and high-grade non-Hodgkin's lymphomas entered the pilot study; 29 were untreated and 7 had relapse disease. Clinical stage was I in 1 patient, II in 7, III in 5 and IV in 23; 10 had B symptoms; the IPI score was 0-2 in 29 cases and > or =3 in the remaining 7. RESULTS: Of the 29 previously untreated patients, 16 achieved complete remission, 8 partial remission, 4 developed progressive disease and 1 was withdrawn early from the study because of acute viral hepatitis; subsequently 4 relapsed and 3 died (2 of disease progression, 1 of causes unrelated to the disease). In the pre-treated group 3 patients obtained complete remission, 2 partial remission and in 1 patient the disease progressed; 3 of these pre-treated patients died (1 of progressive disease, 1 of a new relapse, 1 of myocardial infarction during therapy). With a 20-month median follow-up, the 30-month overall and relapse-free survival were 0.58 and 0.70, respectively. G-CSF was administered to all but 2 patients, with median delivery throughout the whole regimen of 8, 400 microg per patient. Actual cumulative DI was 0.82+/-0.11. Grade 3-4 hematologic toxicity consisted of anemia in 3 cases, of leukopenia in 8 and of thrombocytopenia in 2; the same grade of non-hematologic toxicity involved the liver in 2 cases, the heart in 1 (the above mentioned death), the digestive mucosa in 2 and the peripheral nerves in 1 patient. INTERPRETATION AND CONCLUSIONS: The iso-DI sequential variant of the ProMACE-CytaBOM regimen can be considered feasibile, relatively non-toxic, and can be given on an out-patient basis. Limited use of G-CSF is required (about 3 vials after each drug administration). Thus, a randomized trial with the original ProMACE-CytaBOM regimen can be designed.

Long-term survival of stage I multiple myeloma given chemotherapy just after diagnosis or at progression of the disease: a multicentre randomized study. Cooperative Group of Study and Treatment of Multiple Myeloma.

We conducted a randomized trial to evaluate whether melphalan-prednisone (MPH-P) treatment administered just after diagnosis improves survival of stage I multiple myeloma (MM). Between January 1987 and March 1993, 145 consecutive previously untreated patients with stage I MM were randomized between treatment with MPH-P (administered for 4 days every 6 weeks) just after diagnosis and treatment only at disease progression. Survival was not influenced by MPH-P treatment either administered just after diagnosis or at disease progression (64 vs 71 months respectively). Comparing the first with the second group the odds ratio of death is 1.17 (95% confidence interval 0.57-2.42; P = 0.64). Disease progression occurred within a year in about 50% of patients who were initially untreated. Response rate was similar in both groups, but duration of response was shorter in patients who were treated at disease progression (48 vs 79 months, P = 0.044). Patients actually treated at disease progression (34/70) survived shorter than those who had neither disease progression nor treatment (56 vs > 92 months; P = 0.005). Starting MPH-P just after diagnosis does not improve survival and response rate in stage I MM, with respect to deferring therapy until disease progression. However, patients with stage I MM randomized to have treatment delayed and who actually progressed and were treated had shorter survival than those with stable disease and no treatment. Biologic or other disease features could identify these subgroups of patients.

Circulating progenitor cell release and functional characterization after topotecan plus G-CSF and erythropoietin in small cell lung cancer patients.

Topotecan is a new antineoplastic agent active in ovarian cancer, with promising activity in small cell lung cancer and predictable toxicity. As a part of our ongoing attempt to optimize the use of disease-specific drugs as circulating progenitor cell (CPC) priming in solid tumors, we have evaluated the effects on CPC release of single-agent Topotecan followed by granulocyte colony-stimulating factor (G-CSF) + human recombinant erythropoietin (rhEPO), together with the cell cycle status of the collected CD34+ cells. Ten pretreated patients with small cell lung cancer received Topotecan (1 mg/m2, i.v. for 5 consecutive days) followed by G-CSF (5 microg/kg/day, s.c.) + rhEPO (10,000 I.U. daily, s.c.), starting 24 h after Topotecan. The combination was well tolerated and no relevant side-effects were recorded. On day +10 (range +9 to +11) after the last dose of Topotecan, the median WBC count and the CD34+ cell peak were 8.2 x 10(3) microl (range 4.9-13.9) and 55 microl (range 28-75), respectively. Using flow cytometry, a detailed cell cycle analysis was performed on these CD34+ cells. The cell cycle distribution was determined by DNA content coupled with bromodeoxyuridine incorporation analysis. Apoptosis was evaluated by quantitating DNA strand breaks. The percentage of CD34+ cells in active S-phase was 10.2+/-5%, while early apoptotic CD34+ cells were detected in a low percentage (5.5+/-3%). Topotecan followed by G-CSF + rhEPO mobilizes CPCs effectively. This sequence exerts a stimulation on CD34+ cell cycle with a protective effect from chemotherapy-induced apoptosis. Taken together, these data could be of value for the incorporation of Topotecan, as well as of the combination of G-CSF and rhEPO, into high-dose chemotherapy programs with CPC support.

Cancer chemotherapy-related thrombotic thrombocytopenic purpura: biological evidence of increased nitric oxide production.

The occurrence of thrombotic thrombocytopenic purpura (TTP) in cancer patients receiving chemotherapy has been well established; although this entity is rare, its clinical importance seems to be growing. We describe 3 cases of TTP developing in cancer patients receiving different chemotherapeutic regimens. Using a sensitive high-performance liquid chromatographic method, we evaluated the stable nitric oxide end products, nitrite and nitrate, in the plasma of these patients. Nitric oxide is one of the key components involved in maintaining the normal nonthrombogenicity of the vascular endothelium. In our 3 patients, we found increased nitrate titers that were substantially higher than those observed in patients with de novo TTP. The observed increased release of nitrate could be interpreted as the consequence of massive disruption of endothelial integrity, with consequent passive nitric oxide release in vivo, or an adaptive mechanism of the endothelium to compensate for diffuse microvascular occlusion. The 2 mechanisms may both be involved, but the normal titers of nitric oxide end products in de novo TTP suggest that the former mechanism is more important, at least in cancer chemotherapy-related TTP.

Fetal erythroblast isolation up to purity from cord blood and their culture in vitro.

BACKGROUND: Erythroblasts have been the most encouraging candidate cell type for noninvasive prenatal genetic investigation. We previously showed that human erythroblasts can be recovered from bone marrow and blood bank buffy coats by a physical cell separation. In the present study, we modified our previous methodology, taking into account the peculiar behavior of erythroblasts in response to modifications of pH and osmolality of the separation medium. METHODS: Twenty to forty milliters of cord blood were initially centrifuged on Ficoll/diatrizoate (1.085 g/ml). The interphase cells were further separated on a continuous density gradient (1.040-1.085 g/ml). Two different gradients were initially compared: the first was iso-osmolar and neutral, whereas the second also contained an ionic strength gradient and a pH gradient (triple gradient). A subsequent monocyte depletion was performed by using magnetic microbeads coated with anti-CD14 monoclonal antibody (mAb), and erythroblasts were purified by sedimentation velocity. Purified cells were investigated by analyses with fluorescence-activated cell sorting (FACS) and fluorescence in situ hybridization (FISH) and immunocytochemistry with mAb against fetal hemoglobin and were cultured in vitro. RESULTS: When nucleated cells were spun on an iso-osmolar and neutral continuous density gradient, two separated bands of nucleated red blood cells (NRBCs) were obtained: a light fraction banding at 1.062 g/ml and an heavy fraction banding at 1.078 g/ml. Conversely, when cells were spun in the triple gradient, NRBCs were shifted to the low-density region. Monocyte depletion by immunomagnetic microbeads and velocity sedimentation provided a pure erythroblast population. FACS and FISH analyses and immunocytochemistry substantiated the purity of the isolated cell fraction, which was successfully cultured in vitro. CONCLUSIONS: We have shown that fetal erythroblasts can be purified up to homogeneity from cord blood, but further refinements of the isolation procedure are necessary before the same results can be obtained from maternal peripheral blood.

The degrees of plasma cell clonality and marrow infiltration adversely influence the prognosis of AL amyloidosis patients.

BACKGROUND AND OBJECTIVE: Primary amyloidosis is a lethal form of plasma cell (PC) dyscrasia characterized by deposits of monoclonal immunoglobulin light chains that cause organ dysfunction. In contrast to multiple myeloma, the amyloid clone is typically indolent and of small size, and marrow PC clonality is not always apparent. This is generally investigated by analyzing the light chain isotype ratio in bone marrow PC. We investigated whether the degree of PC infiltration (PC%) and clonality (PC isotype ratio) affected survival in 56 consecutive patients with primary amyloidosis. DESIGN AND METHODS: PC% was determined by morphologic examination. Immunofluorescence microscopy was used to determine the PC light chain isotype ratio. Statistical analysis was carried out using Cox regression models. RESULTS: The degrees of PC clonality and infiltration were inversely correlated with survival (PC isotype ratio, p = 0.001; PC%, p = 0.008). The two variables were weakly correlated (p = 0.02; r = 0.3). Bone marrow PC isotype ratio demonstrated a powerful independent prognostic value at multivariate analysis when analyzed together with congestive heart failure (the major known negative prognostic factor) and PC%. k/l ratio cut-off values of 0.2 (l patients, p = 0.022) and 16 (k patients, p = 0.03) discriminated two groups with a similar number of patients and significantly different survivals. INTERPRETATION AND CONCLUSIONS: PC clonality and marrow infiltration are important parameters that influence prognosis, presumably because they reflect the amount of pathogenic light chain synthesis.

The myelodysplastic syndromes: predictive value of eight prognostic systems in 143 cases from a single institution.

BACKGROUND AND OBJECTIVE: Despite the fact that several prognostic systems for myelodysplastic syndromes (MDS) have been proposed, few studies have been designed to test their effectiveness in independent patient populations. The aim of this study was to compare the prognostic value of 8 previously described prognostic systems in a series of consecutive MDS patients observed at a single institution over a 10-year period. DESIGN AND METHODS: One hundred and forty-three patients were diagnosed as having myelodysplastic syndrome (MDS) according to the French-American-British (FAB) criteria. They were studied retrospectively in order to assess the prognostic value of the FAB classification and 7 other prognostic systems. RESULTS: On the basis of data at diagnosis, all investigated systems effectively stratified patients into groups with different life expectancies and identified a subset of patients with poor clinical outcome. However, the systems had different outcomes concerning median survival of patients classified as low-risk, ranging from less than 3 years for the Mufti scoring system to more than 8 years for the FAB classification modified according to Rosati et al. Moreover, patient distribution into different risk categories was quite different with the different prognostic systems. INTERPRETATION AND CONCLUSIONS: When applied to our case series, some of the prognostic systems had a much lower prognostic value than in the patient population from which they derived. This evidence suggests that testing of prognostic systems in independent case series is necessary before using the systems in clinical practice.