Response to Dr J Parker-Williams

British Journal of Haematology, 2003, 123, 745–753

Correspondence INAPPROPRIATE ANALYSIS OF REPRODUCIBILITY

We read with interest the paper by Giles et al (2003) on the abnormal expression of p53 as a prognostic marker in chronic lymphocytic leukaemia and we would like to make a brief comment. In order to assess the reproducibility of p53 detection by two different pathologists, the authors used Pearson’s correlation coefficient (r) and referred to Table I of their paper, which led them to state that ‘No investigator effect was seen, either in terms of the number of cells deemed positive for p53 expression or in the grading of positivity (Table I). The investigators’ results were highly associated with each other (r ¼ 0Æ980; P < 0Æ001) (Fig 2)’. However, it is our opinion that both r and Table I were insufficient to make such a statement. As Bland & Altman (1986) pointed out in their classic paper, r measures the strength of a relationship between two variables, not the agreement between them. Indeed, we could have perfect correlation (e.g. r value near 1) if the points in Fig 2 laid along any straight line, whereas we would only have perfect agreement if the points in the same figure laid along the line of equality (which they actually did). However, Table I displays the grading of p53 positivity according to each pathologist, but the fact that investigator 1 found a mean of 2Æ5% strongly (3+) positive cells compared with investigator 2, who found a mean of 2Æ9% cells with the same grading, does not mean anything. Indeed, they could be classifying completely different cells as 3+ and this table would not tell us the difference. In summary, although we truly believe that there was a very good agreement between both pathologists in the study

by Giles et al (2003), we think that this paper would benefit from a formal analysis of reproducibility as provided by Bland–Altman technique (Bland & Altman, 1986) or Passing–Bablok regression equation (Passing & Bablok, 1983). Department of Haematology, Hospital Universitario La Paz, Madrid, Spain. E-mail: [email protected]

Julio Delgado Maria Cristina Fernandez-Jimenez

REFERENCES Bland, J.M. & Altman, D.G. (1986) Statistical methods for assessing agreement between two methods of clinical measurement. Lancet, 1, 307–310. Giles, F.J., Bekele, B.N., O’Brien, S., Cortes, J.E., Verstovsek, S., Balerdi, M., Yared, M., Zhou, X., Kantarjian, H.M., Keating, M.J., Thall, P. & Albitar, M. (2003) A prognostic model for survival in chronic lymphocytic leukaemia based on p53 expression. British Journal of Haematology, 121, 578–585. Passing, H. & Bablok, W. (1983) A new biometrical procedure for testing the equality of measurements from two different analytical methods. Application of linear regression procedures for method comparison studies in clinical chemistry, Part I. Journal of Clinical Chemistry and Clinical Biochemistry, 21, 709–720.

Keywords: correlation, coefficient, pearson, agreement, reproducibility.

RESPONSE TO DELGADO AND FERNANDEZ-JIMENEZ We thank Drs Delgado and Fernandez-Jimenez for their comment. We agree that the suggested test is more appropriate for demonstrating agreement and the difference between correlation and agreement should be made known. However, it should be strongly noted that if a correlation is high (i.e. there is a strong linear relationship) and a graphical display of the data show that the data fall along the line of equality, as we showed in Fig. 2 (Giles et al, 2003), then the correlation coefficient and measures mentioned by Drs Delgado and Fernandez-Jimenez will be very similar. Given the obvious relationship between rater scores (evinced by the combined evidence of the correlation
2003 Blackwell Publishing Ltd

and the figure) other statistics, which are not more informative, are at best interesting exercises but do not help the typical medical reader understand the methods used. University of Texas, MD Anderson Cancer Center Houston, TX, USA. E-mail: maher.x.albitar@ questdiagnostics.com

B. Nebiyou Bekele Francis J. Giles Maher Albitar

745

746

Correspondence

REFERENCE Giles, F.J., Bekele, B.N., O’Brien, S., Cortes, J.E., Verstovsek, S., Balerdi, M., Yared, M., Zhou, X., Kantarjian, H.M., Keating, M.J., Thall, P. & Albitar, M. (2003) A prognostic model for survival in

chronic lymphocytic leukaemia based on p53 expression. British Journal of Haematology, 121, 578–585.

Keywords: statistic evaluation, correlation, agreement.

SHORT PROGRESSION-FREE SURVIVAL IN MYELOMA PATIENTS RECEIVING RITUXIMAB AS MAINTENANCE THERAPY AFTER AUTOLOGOUS TRANSPLANTATION We read with interest the case of a patient with multiple myeloma (MM) who responded to the anti-CD20 monoclonal antibody, rituximab (Hofer et al, 2003). Indeed, myeloma cells either lack CD20 antigen or express it very weakly in the large majority of patients (Treon & Anderson, 2000). Thus, although occasional clinical responses have been reported in selected patients with CD20+ myelomatous plasma cells (Treon et al, 2002), MM is usually not considered as a disease suitable for anti-CD20 therapy. However, it has been also suggested that circulating CD20+ clonotypic B cells could act as precursors or ‘neoplastic stem cells’ in MM patients, representing the proliferative compartment of the disease able to play a role in determining relapse after effective treatments (Treon et al, 2002). On this basis, we started a pilot clinical trial in September 2000 that aimed to explore the potential efficacy of rituximab as part of a maintenance therapy programme in MM patients undergoing autologous peripheral blood stem cell transplantation (PBSCT). By July 2001, six consecutive patients with CD20 negative, untreated stage IIIA MM (two males, four females, aged 48–60 years) had been enrolled in an intensive sequential chemotherapy protocol consisting of three monthly courses of VMD (vincristine, mitoxantrone and dexamethasone), followed by three consecutive noncross-resistant mobilizing regimens including high dose cyclophosphamide, etoposide, cisplatinum, cytosine arabinoside, carmustine (BCNU), melphalan and steroids plus granulocyte colony-stimulating factor. Circulating CD34+ autologous stem cells were successfully collected after at least two of these regimens. High-dose therapy with melphalan 140 mg/m2 + fractioned TBI (1200 cGy) or with melphalan 200 mg/m2 was subsequently performed, followed by infusion of a median of 6Æ9 · 106/kg CD34+ cells (range 3Æ2–12Æ8). The median interval between diagnosis and PBSCT was 6.2 months. Two patients achieved complete remission, showing negative immunofixation and marrow plasma cells 50–75%. Three months later, after obtaining local Ethical Committee approval and patient informed consent, these patients received four weekly infusions of anti-CD20 monoclonal antibody (Mabthera, Roche), each at the dose of 375 mg/m2, followed by maintenance therapy with a-interferon (Intron-A; Schering-Plough), at the dose of 3 MU s.c. t.i.w., until progression. As of July 2003, five of these patients were alive, but all of them had relapsed 4–15 months (median 11 months) after PBSCT. As the progression-free survival in our previous series of 21 MM

undergoing PBSCT (using the same protocol, but without anti-CD20 antibody) was 38 months (unpublished observations), we considered these results unacceptable and the trial was stopped. Salvage therapy with single or multiple additional PBSCT (using a melphalan dose of 100 mg/m2) or with a combination of thalidomide, dexamethasone and zoledronic acid, enabled a second remission to be achieved in all cases. Although the number of patients is too low to draw definitive conclusions, in our experience the use of rituximab as maintenance therapy after PBSCT was associated with an unexpectedly high rate of early relapse in MM patients. Recently, a case of plasma cell leukaemia treated with autologous transplantation and early relapse after consolidation therapy with anti-CD20 antibody, has been reported (Gemmel et al, 2002). The reasons of these findings are not clear. We hypothesize a possible role for rituximab in provoking a further decrease in the residual, normal B-cell activity within the context of the complex immunological network characterizing anti-tumour immune response. The therapeutic role of rituximab in MM remains elusive and requires ad hoc studies. Unit of Hematology and Stem Cell Transplantation, IRCCS ‘‘Casa Sollievo della Sofferenza’’ Hospital, San Giovanni Rotondo, Italy. E-mail: [email protected]

Pellegrino Musto Angelo Michele Carella Jr Michele Mario Greco Antonietta Falcone Grazia Sanpaolo Carlo Bodenizza Nicola Cascavilla Lorella Melillo Angelo Michele Carella

REFERENCES Gemmel, C., Cremer, F.W., Weis, M., Witzens, M., Moldenhauer, G., Koniczek, K.H., Imbach, U., Ho, A.D., Moos, M. & Goldschmidt, H. (2002) Anti CD-20 antibody as consolidation therapy in a patients with primary plasma cell leukemia after high-dose therapy and autologous stem cell transplantation. Annals of Hematology, 81, 119–123. Hofer, S., Hunziker, S., Dirnhofer, S. & Ludwig, C. (2003) Rituximab effective in a patient with refractory autoimmune haemolytic


2003 Blackwell Publishing Ltd, British Journal of Haematology 123: 745–753

Correspondence anaemia and CD20-negative multiple myeloma. British Journal of Haematology, 122, 690–691. Treon, S.P. & Anderson, K.C. (2000) The use of rituximab in the treatment of malignant and non-malignant plasma cell disorders. Seminars in Oncology, 27, 79–85. Treon, S.P., Pilarski, L.M., Belch, A.R., Kelliher, A., Preffer, F.I., Shima, Y., Mitsiades, C.S., Mitsiades, N.S., Szczepek, A.J., Ellman, L., Harmon, D., Grossbard, M.L. & Anderson, K.C. (2002) CD-20

747

directed serotherapy in patients with multiple myeloma: biologic considerations and therapeutic applications. Journal of Immunotherapy, 25, 72–81.

Keywords: multiple myeloma, anti-CD20, autologous stem cell transplantation.

rituximab,

SERUM MUC-1 AS A MARKER OF DISEASE STATUS IN MULTIPLE MYELOMA PATIENTS RECEIVING THALIDOMIDE We read with interest the recent publication (Luminari et al, 2003) regarding the assessment of serum mucin-1 (sMUC-1) levels in plasma cell dyscrasias. Their study involved an analysis of consecutive patients referred to their Institute. Although not stated, we assume that the subgroup of multiple myeloma (MM) patients would have been heterogeneous in terms of amount of prior therapy and subsequent treatment. One of their conclusions was that elevated sMUC-1 levels correlated with a shorter survival in MM patients. We too have been interested in the value of sMUC-1 as a potential prognostic indicator but have not been able to demonstrate the same finding. We prospectively measured sMUC-1 in patients with relapsed/refractory MM as part of a recently published phase-II clinical trial of thalidomide therapy (Mileshkin et al, 2003). sMUC-1 levels were measured at baseline in 62 of the 75 patients using the same method as Luminari et al (2003) (Chiron diagnostics: upper limit of normal ¼ 31 U/ml). Of note, sMUC-1 levels were elevated in 35% of our patients, a substantially higher proportion than that observed by Luminari et al (2003) (17%). Nonetheless, the median sMUC-1 level of 46 U/ml (range 32–158 U/ml) that we observed was very similar to their analysis (43Æ2 U/ml). The reason for our observation of a higher frequency of elevated sMUC-1 is unclear but may relate to differences in the populations studied. Like their study, we were unable to demonstrate a correlation between sMUC-1 levels and various parameters of MM, namely bone marrow plasma cell infiltrate, serum/ urine paraprotein level, immunoglobulin or light chain subtype, b2-microglobulin level, C-reactive protein, platelet count or serum creatinine. Furthermore, we were unable to confirm their observation of a correlation between elevated sMUC-1 and elevated lactate dehydrogenase or anaemia. Similar to Treon et al (2000), we demonstrated that patients with more advanced disease were most likely to have an elevated baseline sMUC-1 [stage 1, zero of three patients (0%); stage 2, 14 of 39 (36%) and stage 3, nine of 20 (45%); P ¼ 0Æ19]. A subset (n ¼ 36) of our patients had bone marrow biopsy sections stained for MUC-1 expression by immunohistochemistry (IHC) using a mouse monoclonal antibody (BC2). Serum MUC-1 levels correlated with the bone marrow IHC score

(% positive plasma cells · intensity): R ¼ 0Æ43 (P ¼ 0Æ047). However, six patients (27%) with elevated sMUC-1 had negligible marrow plasma cell staining for MUC-1. In univariate analysis, we found a trend for a higher response rate (RR) among patients with normal levels of sMUC-1 (RR 35% versus 14% for those with elevated levels: P ¼ 0Æ084). However, in multivariate analysis, sMUC-1 was not predictive of response. At 18 months median follow-up we could find no difference between patients with normal versus elevated sMUC-1 levels in terms of progression-free survival (median 6Æ1 months versus 5 months, P ¼ 0Æ31) or overall survival (15 months versus 16 months, P ¼ 0Æ25). In summary, our results in a group of heavily pretreated patients managed in a homogeneous fashion contradicts that of Luminari et al (2003), as we have been unable to demonstrate a relationship between sMUC-1 and survival. Nonetheless, we believe that the value of sMUC-1 levels does require further study. Newer techniques such as array analysis may give additional insights into its role in disease pathogenesis. To fully evaluate the role of sMUC-1 as a prognostic marker, studies with larger patient numbers are required, particularly examining patients with previously untreated disease. We also continue to explore the relationship between sMUC-1 levels and marrow plasma cell MUC-1 expression, and the role of sMUC-1 as a measure of response to therapy. Division of Haematology and Medical Oncology, Peter MacCallum Cancer Centre, Melbourne, Victoria, Australia. E-mail: [email protected]

Linda Mileshkin H. M. Prince J. F. Seymour J. J. Biagi

REFERENCES Luminari, S., Goldaniga, M., Ceccherelli, F., Guffanti, A., Bombardieri, E., Marcheselli, R., Cro, L., Colombi, M., Federico, M. & Baldini, L. (2003) Prevalence and prognostic significance of sMUC-1 levels in plasma cell dyscrasias. British Journal of Haemotology, 121, 772–774. Mileshkin, L., Biagi, J.J., Mitchell, P., Underhill, C., Grigg, A., Bell, R., McKendrick, J., Briggs, P., Seymour, J.F., Lillie, K., Smith, J.G.,


2003 Blackwell Publishing Ltd, British Journal of Haematology 123: 745–753

748

Correspondence

Zeldis, J.B. & Prince, H.M. (2003) Multicenter phase 2 trial of thalidomide in relapsed/refractory multiple myeloma: adverse prognostic impact of advanced age. Blood, 102, 69–77. Treon, S.P., Maimonis, P., Bua, D., Young, G., Raje, N., Mollick, J., Chauhan, D., Tai, Y.T., Hideshima, T., Shima, Y., Hilgers, J., von Mensdorff-Pouilly, S., Belch, E.R., Pilarski, L.M. & Anderson, K.

(2000) Elevated soluble MUC-1 levels and decreased anti-MUC-1 antibody levels in patients with multiple myeloma. Blood, 96, 3147–3153.

Keywords: multiple myeloma, MUC-1, prognostic factor.

SERUM MUC-1 AS A MARKER OF DISEASE STATUS IN MULTIPLE MYELOMA PATIENTS RECEIVING THALIDOMIDE – RESPONSE TO MILESHKIN ET AL. Different from our recently published study (Luminari et al, 2003), Dr Mileshkin and colleagues could not confirm our conclusions on the prognostic role of elevated serum mucin1 (sMUC-1) levels in patients with multiple myeloma (MM) and were also unable to confirm our finding of a correlation between sMUC-1 levels and elevated lactate dehydrogenase or anaemia in their study population. Moreover, if the median sMUC-1 concentration was similar (46 U/ml vs. 43.2 U/ml of our series), they found a higher prevalence of patients with elevated sMUC-1 levels (35% vs. 13% of our series). As previously demonstrated by Treon et al (2000), they could confirm a trend towards a higher prevalence of sMUC-1 levels in patients with advanced stage of disease. The different study population may be assumed as the main reason for the different conclusions. In our study we enrolled only consecutive patients with de novo, untreated, MM and plasma cell leukaemia (PCL) while Mileshkin’s patients were either relapsed or refractory subjects and no PCL was included. Little is known about MUC-1 biology and its behaviour during the different phases of disease, but dexamethasone has been shown to induce MUC-1 overexpression (Treon et al, 1999); previous treatments could therefore represent a bias when evaluating sMUC1 in a population of heavily pretreated patients. Moreover, according to our manufacturer’s guidelines (Immulite, Los Angeles, CA, USA) the cut-off we used to define patients as having elevated sMUC-1 levels was 51 U/ ml, higher than that used by Mileshkin (31 U/ml) and this could also explain the lower prevalence seen in our patients if a similar median concentration of sMUC-1 was observed. Although Mileshkin and colleagues could not confirm our conclusions in their series of pretreated patients, their data, along with ours, confirm some role of MUC-1 in the biology of plasma cell disorders that should be further investigated in larger series of patients. Apart from overt disease one of

the most relevant findings that should be addressed is the biological and prognostic meaning of elevated sMUC-1 levels in subjects with monoclonal gammopathy of undetermined significance: this represents one of the first evidences of an association between a tumour marker and a benign condition. 1 Dipartimento di Oncologia ed Ematologia, Universita` di Modena e Reggio Emilia, Modena, 2Dipartimento di Ematologia, Ospedale Maggiore IRCCS, Milano, Italy. E-mail: [email protected]

Stefano Luminari 1 Massimo Federico 1 Luca Baldini 2

REFERENCES Luminari, S., Goldaniga, M., Ceccherelli, F., Guffanti, A., Bombardieri, E., Marcheselli, R., Cro, L., Colombi, M., Federico, M. & Baldini, L. (2003) Prevalence and prognostic significance of sMUC-1 levels in plasma cell dyscrasias. British Journal of Haematology, 121, 772–774. Treon, S.P., Mollick, J.A., Urashima, M., Teoh, G., Chauhan, D., Ogata, A., Raje, N., Hilgers, J.H., Nadler, L., Belch, A.R., Pilarski, L.M. & Anderson, K.C. (1999) Muc-1 core protein is expressed on multiple myeloma cells and is induced by dexamethasone. Blood, 93, 1287–1298. Treon, S.P., Maimonis, P., Bua, D., Young, G., Raje, N., Mollick, J., Chauhan, D., Tai, Y.T., Hideshima, T., Shima, Y., Hilgers, J., von Mensdorff-Pouilly, S., Belch, A.R., Pilarski, L.M. & Anderson, K.C. (2000) Elevated soluble MUC1 levels and decreased anti-MUC1 antibody levels in patients with multiple myeloma. Blood, 96, 3147–3153.

Keywords: multiple myeloma, plasma cell disorders, sMUC1 levels.

CONSTITUTIONAL PERICENTRIC INVERSION OF CHROMOSOME 9 AND BONE MARROW TRANSPLANTATION In response to our initial observation of delayed engraftment in two bone marrow transplantation patients from unrelated donors with constitutional inv(9) (Keung et al, 2002), Imashuku et al (2002) reported a case of autologous peripheral blood stem cell transplantation and delayed

engraftment in a patient with acute myeloid leukaemia (AML) in second remission and constitutional inv(9). We describe another patient with AML and inv(9), similar to the case of Imashuku et al (2002), who subsequently underwent allogeneic bone marrow transplantation from a


2003 Blackwell Publishing Ltd, British Journal of Haematology 123: 745–753

Correspondence normal unrelated donor instead of autologous transplantation. In this case, there was no engraftment problem. CASE HISTORY A 26-year-old female was diagnosed with AML, French– American–British (FAB) classification type M2, a few days after normal spontaneous vaginal delivery of a healthy male infant. Cytogenetic study of the bone marrow revealed 46 XX, t(8;15)(q24.1;q15), inv(9)(p11q13) in all 20 metaphases analysed. The patient received induction chemotherapy with idarubicin, etoposide and cytarabine. Complete remission was readily obtained and she received two cycles of consolidation chemotherapy. Three months after the last chemotherapy, she remained pancytopenic and required weekly blood and platelet transfusions. Bone marrow biopsy showed severely hypocellular marrow with no evidence of leukaemic recurrence. Cytogenetic study revealed 46 XX, inv(9)(p11q13) [5]. Five months after the last chemotherapy, the patient remained pancytopenic and was referred to our institution for allogeneic bone marrow transplantation. Her white blood cell count was 1Æ1 · 109/l, haemoglobin 9Æ4 g/dl, mean cell colume 80Æ4 fl, platelet count 25 · 109/l. Repeat bone marrow biopsy showed severely hypocellular marrow without leukaemic recurrence. Cytogenetic study of the bone marrow revealed female phenotype of 46 XX, inv(9)(p11q13). Four months later, she underwent allogeneic bone marrow transplantation, nucleated cell count 2Æ56 · 108/kg and CD34+ cells 1Æ74 · 106/kg, from a human leucocyte antigen-matched unrelated male donor. The conditioning regimen consisted of oral busulphan 1 mg/kg every 6 h for 16 doses and cyclophosphamide 60 mg/kg/d for two doses. Myeloid and platelet engraftment was achieved on days +14 and +31, respectively. Repeat bone marrow examination at 3 months was normal with

749

complete donor male phenotype of 46 XY. She remains healthy 2 years following transplant. Previous case reports of delayed engraftment after autologous and allogeneic transplantation of inv(9) donors (Imashuku et al, 2002; Keung et al, 2002) suggest impaired engraftment potential of the inv(9) haematopoietic stem cells. The failure of recovery from pancytopenia induced by consolidation chemotherapy in this patient with inv(9), followed by uneventful engraftment after transplant from a normal donor further supports this hypothesis. Further studies are required to confirm this observation and hypothesis. 1

Section on Hematology and Oncology, and 2Medical Genetics Laboratory, Wake Forest University Comprehensive Cancer Center, Winston-Salem, NC, USA. E-mail: [email protected]

Yi-Kong Keung 1 Mary Ann Knovich 1 David D. Hurd 1 Mark Pettenati, 2

REFERENCES Imashuku, S., Naya, M., An, B., Nakabayashi, Y., Kuriyama, K., Udeda, I., Morimoto, A., Hibi, S. & Todo, S. (2002) Constitutional pericentric inversion of chromosome 9 and haemopoietic stem cell transplantation: delayed engraftment. British Journal of Haematology, 118, 1195–1196. Keung, Y.K., Pettenati, M.J., Hurd, D.D., Powell, B.L. & Buss, D.H. (2002) Allogeneic marrow grafts from donors with congenital pericentric inversion of chromosome 9. British Journal of Haematology, 116, 238–239.

Keywords: pericentric inversion, chromosome 9, acute myeloid leukaemia, inv(9).

c-FMS MUTATIONAL ANALYSIS IN ACUTE MYELOID LEUKAEMIA The human c-FMS, located on chromosome 5 at band 5q33.3, encodes a 972 amino acid transmembrane glycoprotein which functions as the receptor of the monocytecolony stimulating factor (M-CSF or CSF-1). Early studies suggested a leukaemogenic role, as overexpression of c-FMS in mice leads to the development of myeloblastic leukaemia. In addition, studies employing hybridization to allele-specific oligonucleotide (ASO) probes identified activating point mutations in approximately 18% of acute myeloid leukaemia (AML) and 15% of myelodysplastic syndrome (MDS) patients (Ridge et al, 1990; Tobal et al, 1990), with codon 969 being more frequently involved than codon 301. Mutations were more common in leukaemias characterized by monocytic differentiation, namely acute myelomonocytic leukaemia (M4-AML) and chronic myelomonocytic leukaemia (CMML). However, Shepherd et al (1990), utilizing direct sequencing, failed to find codon 969 and 301 mutations in either AML or MDS, a finding that has been supported by subsequent studies (Springall et al, 1993;

Misawa et al, 1997). The role of c-FMS mutations in leukaemogenesis, therefore, remains controversial and the explanation for the discrepancy may lie in the fact that the early studies used ASO hybridization techniques where probe specificity, if not carefully controlled, may yield misleading results. In order to clarify the pathogenetic relevance of c-FMS mutations in AML, we have screened, for the first time, the entire coding sequence and intron/exon boundaries of exons 2–22 of the gene, using conformation sensitive gel electrophoresis (CSGE) as previously described (Abu-Duhier et al, 2003). Genomic DNA was obtained from the marrow at presentation of 60 cases of AML entered into the Medical Research Council (MRC) AML X trial (32 males, 28 females, mean age 41.6 years, range 15–59 years). The cases were classified, according to the French–American–British (FAB) criteria, as M0 (n ¼ 4), M1 (n ¼ 8), M2 (n ¼ 10), M3 (n ¼ 10), M4 (n ¼ 12), M5 (n ¼ 10) and M6 (n ¼ 6). Standard cytogenetic analysis demonstrated inv(16)


2003 Blackwell Publishing Ltd, British Journal of Haematology 123: 745–753

750

Correspondence

(n ¼ 7), t(8;21) (n ¼ 2), t(15;17) (n ¼ 10), other abnormalities (n ¼ 25) and a normal karyotype (n ¼ 16). Genomic DNA was also prepared from the peripheral blood of 70 normal individuals using the Nucleon Biosciences BACC II kit. Genomic DNA was amplified using the polymerase chain reaction. Three patients were found to have mutations, including two with a novel exon 6 mutation (nucleotide 10906 G > T) that is predicted to result in an alanine to serine substitution at codon 245. These mutations were detected in a 52-year-old male (46XY, inv(16)(p13;q22), FAB group M1) and a 49-year-old male (46XY, FAB group M6). Interestingly, both cases lacked FLT3 (internal tandem duplication and Asp835) and c-KIT (exon 8 and Asp816) mutations. The third case, a 32-year-old male with acute promyelocytic leukaemia, possessed an exon 9 mutation (nucleotide 18073 G > A) that is predicted to result in a glycine to serine substitution at codon 413. The latter case also possessed an FLT3 Asp835 mutation. The Gly413Ser change has recently been reported in 5% of patients with idiopathic myelofibrosis (Abu-Duhier et al, 2003). Neither of these changes was detected in 70 normal control subjects. We were not able to detect the previously reported c-FMS codon 301 and 969 mutations. Therefore, our data suggests that c-FMS codon 301 and 969 mutations are rare events in AML and that, as a result, they are unlikely to play a significant role in leukaemogenesis. This conclusion is in agreement with recent studies that employed similar screening methods (Misawa et al, 1997; Meshinchi et al, 2003). Using the same techniques, we have previously identified several mutations in the c-KIT and FLT3 genes (Abu-Duhier et al, 2003). However, in this study, we have identified novel c-FMS mutations involving exons 6 and 9 in a small number of AML patients. The biological consequences of these changes remain unclear and further study is required. Nevertheless, both mutations are located in the extracellular domain of c-FMS, with Gly413 being a highly conserved amino acid. We have previously reported mutations affecting the nearby residue Asp419 of c-KIT, a fact that highlights the likely importance of this region of class III receptor tyrosine kinases (RTKs). It is possible that these changes could lead to constitutive activation of the receptor, as random mutagenesis of c-FMS has revealed multiple sites for activating mutations within the extracellular domain. Finally, mutually exclusive RTK class III mutations appear to be characteristic of AML with

inv(16), occurring in approximately 40% of cases (Care et al, 2003). It is of interest, therefore, that the c-FMS exon 6 mutation that occurred in the AML with inv(16) was not associated with either an FLT3 or c-KIT mutation. Academic Unit of Haematology, Division of Genomic Medicine, Royal Hallamshire Hospital, Sheffield, UK. E-mail: [email protected]

Faisel M. Abu-Duhier Anne C. Goodeve Giu A. Wilson Ian R. Peake John T. Reilly

REFERENCES Abu-Duhier, F.M., Goodeve, A.C., Care, R.S., Gari, M., Wilson, G., Peake, I. R. & Reilly, J.T. (2003) Mutational analysis of class III receptor tyrosine kinase genes (c-KIT, c-FMS, FLT3) in idiopathic myelofibrosis. British Journal of Haematology, 120, 464-470. Care, R.S., Valk, P.J.M., Goodeve, A.C., Abu-Duhier, F.M., GeertsmaKleinekoort, W.M.C., Wilson, G.A., Gari, M.A., Peake, I.R., Lo¨wenberg, B. & Reilly, J.T. (2003) Incidence and prognosis of c-KIT and FLT3 mutations in core binding factor (CBF) acute myeloid leukaemias. British Journal of Haematology, 121, 775–777. Meshinchi, S., Stirewalt, D.L., Alonzo, T.A., Zhang, Q., Sweetser, D.A., Woods, W.G., Bernstein, I.D., Arceci, R.J. & Radich, J.P. (2003) Activating mutations of RTK/ras signal transduction pathway in pediatric acute myeloid leukaemia. Blood (in press). Misawa, S., Horiike, S., Kaneko, H. & Kashima, K. (1997) Genetic aberrations in the development and subsequent progression of myelodysplastic syndrome. Leukemia, 11(Suppl. 3), 533–535. Ridge, S.A., Worwood, M., Oscier, D., Jacobs, A. & Padua, R.A. (1990) FMS mutations in myelodysplastic, leukemic, and normal subjects. Proceedings of the National Academy of Sciences of the United States of America, 87, 1377–1380. Shepherd, L., Cameron, C., Graham, C., Hoyle, S., Galbraith, P.R. & Lillicrap, D. (1990) Direct sequencing of the N-ras, c-FMS and p53 loci for transforming mutations in myelomonocytic leukaemia. Blood, 76(Suppl. 1), 245a. Springall, F., O’Mara, S., Shounan, Y., Todd, A., Ford, D. & Illand, H. (1993) c-FMS point mutations in acute myeloid leukaemia: fact or fiction? Leukemia, 7, 978–985. Tobal, K., Pagliuca, A., Bhatt, B., Bailey, N., Layton, D.M. & Mufti, G.J. (1990) Mutation of the human FMS gene (M-CSF receptor) in myelodysplastic syndromes and acute myeloid leukemia. Leukemia, 4, 486–489.

Keywords: c-FMS, RTK Class III, AML.

IMMUNOPLATELET COUNTING: PLATELET TRANSFUSIONS In their timely assessment of the need to improve the platelet counting ability of auto-analysers, especially at the low levels required for making clinical decisions on whether to offer patients a platelet transfusion, Norris et al (2003) were rightly critical that the United Kingdom National External Quality Assessment Scheme for Haematology [UK NEQAS (H)] was not providing survey samples with low platelet counts, specifically less than 10 · 109/l. At the time of writing, this comment was correct. The UK NEQAS (H) had

no difficulty in providing survey samples with platelet numbers down to this level. However, without a reliable technique of validating the low platelet count, we were unwilling to offer external quality assessment (EQA) material that we could not vouch for. The penalty is that in lowering the count, the coefficient of variance (CV) increases. The counting technology of these automated blood count analysers has improved, although there is still some debate as to which technology is better. Clearly, there are differences


2003 Blackwell Publishing Ltd, British Journal of Haematology 123: 745–753

Correspondence Table I. Mean values and percentage coefficient of variance (CV%) for platelet counts (109/l) returned on two the United Kingdom National External Quality Assessment Scheme for Haematology [UK NEQAS (H)] full blood count survey samples.

Sample 1 Instruments

Sample 2

No. tested Mean CV% Mean

Abbott Cell-Dyn 3500 46 Abbott Cell-Dyn 4000 67 ABX Pentra 120 54 Bayer ADVIA 120 164 Beckman Coulter Gen S 139 Beckman Coulter Max M 47 Beckman Coulter STKS 62 Sysmex SF3000 51 Sysmex XE2100 89

5Æ6 10Æ5 4Æ5 5Æ7 4Æ4 2Æ5 5Æ1 2Æ8 4Æ4

47Æ8 22Æ5 66Æ2 29Æ7 35Æ4 50Æ4 23Æ1 53Æ7 )

CV%

240Æ4 11Æ0 215Æ5 6Æ6 241Æ9 5Æ0 260Æ5 5Æ7 242Æ0 6Æ8 228Æ5 7Æ6 226Æ7 9Æ7 230Æ5 5Æ3 232Æ0 3Æ7

between the Instrument groups, and Table I illustrates the problem. The two platelet counts were taken from the same survey (Survey 0210FB, unpublished observations), and show what can be expected across a wide range of instruments (only those with over 45 instruments in the group are shown). What is certain is that with some 1250 instruments participating in the scheme, and, however, good individual results may be, it is unlikely that the CV of less than 3%, attained by the immunoplatelet method, will be matched. There is always room for improvement, but UK NEQAS (H) does not wish to penalize laboratories for the limitations in instrument technology that is outside their control. The objective will be to continue to monitor the ability of the instruments to count platelets at levels where clinical decisions have to be made, and that this is carried out repeatedly and reliably.

751

Over the past year, the UK NEQAS (H) has been using the immunoplatelet count to measure the platelet count of our survey material. We are now comfortable that low platelet counts are what we believe them to be. I am certain that Dr Norris and colleagues have noticed that a number of our blood count distributions have had low platelet counts, even down to 5 · 109/l in a survey earlier this year. If I understand the authors correctly, this level may in time become the target for making a clinical decision, in the absence of bleeding manifestations, on the necessity for a platelet transfusion. I suspect we have reached the limit of technology with the instruments that are currently available for platelet counts down to the 5 · 109/l level. It seems likely that a clinical decision to transfuse platelets can be made with some confidence, taking into account the platelet count figures returned to the UK NEQAS (H) from the 1250 instruments registered in the scheme. It may still be wise to examine a stained blood film first. Department of Haematology, St George’s Hospital, Blackshaw Road, London, UK. E-mail: [email protected]

J. Parker-Williams

REFERENCE Norris, S., Pantelidou, D., Smith, D. & Murphy, M.F. (2003) Immunoplatelet counting: potential for reducing the use of platelet transfusions through more accurate platelet counting. British Journal of Haematology, 121, 605–613.

Keywords: platelet counting, United Kingdom National External Quality Assessment Scheme for Haematology , platelet transfusions.

RESPONSE TO DR J PARKER-WILLIAMS In reply to Dr Parker-Williams letter, we acknowledge and welcome the recent distribution of low platelet counts by the United Kingdom National External Quality Assessment Scheme (UK NEQAS). We were unaware at the time of submission of the use of the immunoplatelet count by the scheme and also welcome its use. We would also like to respond to the point made by Dr Parker-Williams that the 5 · 109/l platelet transfusion threshold may, in time, become the target for clinical decision making. As stated in our paper (Norris et al, 2003), the 1997 Consensus Conference on Platelet Transfusion, Edinburgh (Consensus Conference on Platelet Transfusion, 1997) recommended that for uncomplicated patients, evidence on the safety of thresholds for prophylactic platelet transfusions lower than 10 · 109/l should be sought, but recognized that problems in accurate counting of platelets below 10 · 109/l may make it difficult to do this. The recent guideline from the British Committee for Standards in

Haematology (2003) for the use of platelet transfusions include the following recommendations: 1. Guidelines for indications for platelet transfusions A threshold of 10 · 109/l is as safe as higher levels for patients without additional risk factors. Risk factors include sepsis, concurrent use of antibiotics or other abnormalities of haemostasis (Grade A, Level Ib). For patients without any risk factors a threshold of 5 · 109/l may be appropriate if there are concerns that alloimmunisation could lead to platelet refractoriness (Grade B, Level IIa). However, accurate counting of low platelet numbers may create difficulties when trying to reduce the threshold below 10 · 109/l. 2. Approaches for minimizing the use of platelet transfusion Lowering of platelet threshold from 20 · 109/l to 10 · 109/ l: there is the possibility of further reduction to 5 · 109/l for prophylactic platelet transfusion, but this would require the


2003 Blackwell Publishing Ltd, British Journal of Haematology 123: 745–753

752

Correspondence

routine availability of methods for accurate platelet counting. We agree that the optimal threshold for prophylactic platelet transfusions remains uncertain, and consider that clinical trials of lower thresholds are not possible while routine haematology analysers are unreliable for platelet counting of low counts. In support of this, we have further, as yet unpublished, data from a joint study with Professor Machin’s group at University College, London and colleagues from a number of other large haematology centres in the UK that auto-analysers give increasingly unreliable results compared with immunoplatelet counting as the platelet count falls below 20 · 109/l. These findings suggest that the next step in determining the most effective approach to the use of platelet transfusions in haematology should be to revisit the question of prophylactic versus therapeutic transfusions in a clinical trial. Dr Parker-Williams indicates that the low platelet count distributions produce coefficients of variance (CVs) that do not reflect the fact that the majority of auto-analysers were giving fairly reasonable results. We agree that it will be practically impossible for a National External Quality Assurance Scheme to achieve respectable CVs at low platelet counts given the large number of diverse analysers in such schemes. Indeed, our paper suggested that there is probably a large variation in reported platelet counts at these low levels, at a national level. We do suggest that, despite the participants dislike of the CVs that they may

return, schemes such as NEQAS have a responsibility to produce quality control material at levels of platelet counts where clinical decisions are being made. 1

Department of Haematology, Oxford Radcliffe NHS Trust, and 2 National Blood Service, Oxford Centre, The John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK. E-mail: [email protected]

Scott Norris 1 Despina Pantelidou 1 DanSmith 1 MichaelMurphy 1,2

REFERENCES British Committee for Standards in Haematology (2003) Guidelines for the use of platelet transfusions. British Journal of Haematology, 122, 10–23. Consensus Conference on Platelet Transfusion (1997) Synopsis of background papers and consensus statement. British Journal of Haematology, 101, 609–617. Norris, S., Pantelidou, D., Smith, D. & Murphy, M.F. (2003) Immunoplatelet counting: potential for reducing the use of platelet transfusions through more accurate platelet counting. British Journal of Haematology, 121, 605–613.

Keywords: platelets, immuno-platelet count, platelet transfusion, platelet counting, platelet transfusion threshold.

LATE OCCURRENCE OF HEPATIC VENO-OCCLUSIVE DISEASE FOLLOWING GEMTUZUMAB OZOGAMICIN: SUCCESSFUL TREATMENT WITH DEFIBROTIDE Gemtuzumab ozogamicin (GO; Mylotarg CMA-676) is a novel chemotherapeutic agent consisting of an anti-CD33 monoclonal antibody linked to calicheamicin (Zein et al, 1988). The myeloid cell surface antigen CD33 is expressed by the leukaemic blasts of at least 90% of acute myeloid leukaemia (AML) patients. CD33-positive AML patients in first relapse treated with GO have been reported to achieve a 20–30% response rate (Sievers et al, 2001; Larson et al, 2002a). A possible adverse event associated with Mylotarg is hepatotoxicity, with 20% of patients developing grade 3 and 4 hyperbilirubinaemia, and 17% developing grade 3 and 4 elevations in the transaminases [aspartate transaminase (AST), alanine transaminase (ALT)] in the reported phase II studies (Giles et al, 2001). Hepatic veno-occlusive disease (VOD) is a clinical syndrome diagnosed by Seattle and Baltimore standard criteria, including hyperbilirubinaemia (bilirubin > 34Æ2 lmol/l) accompanied by one or more of the following symptoms: painful hepatomegaly, fluid retention (e.g. ascites), or sudden weight gain (>5% of pretreatment weight) (Giles et al, 2001). In three phase II studies, the incidence of hepatic VOD/ sinusoidal obstruction syndrome (SOS) was 3% (seven of 277) when GO was administered to 277 patients with AML in first relapse, at a dose of 9 mg/m2 i.v. for two doses, 14 d

apart. The incidence of fatal VOD/SOS was 1% (four of 277) (Larson et al, 2002b). The incidence of VOD in association with GO treatment was higher than 3% in some postmarketing experience in patients previously treated with standard chemotherapy or haematopoietic stem cell transplant (Giles et al, 2001; Cohen et al, 2002). In a retrospective study the incidence of VOD was 64% (nine of 14) when GO was administered to patients with AML, before undergoing myeloablative allogeneic stem cell transplantation (Wadleigh et al, 2003). The median time of occurrence of VOD was 25 d (range, 10–35 d) after the first infusion of GO (Giles et al, 2001). We report the unusual case of a 67-year-old female patient with AML, refractory to standard induction therapy, who developed VOD after GO treatment. At the beginning of GO therapy, her bilirubin level was 15Æ4 lmol/l, while AST and ALT were 24 and 33 U/l respectively. At discharge, on day +42 after GO therapy, liver enzymes and bilirubin were normal and maintained within the normal range until day +69, when the patient was referred again to the hospital for abdominal pain. The physical examination revealed liver enlargement, ascites and weight gain (8 kg in 7 d). Laboratory tests showed hyperbilirubinaemia (bilirubin 59Æ8 lmol/l) without any increase of the transaminases,


2003 Blackwell Publishing Ltd, British Journal of Haematology 123: 745–753

Correspondence grade 2 leucopenia and grade 4 thrombocytopenia. An ultrasound echotomography performed on day +70 confirmed ascites, showing findings highly suggestive of VOD (increased hepatic artery resistance index and decreased portal venous flow). Because of severe thrombocytopenia, the patient was treated with defibrotide at 20 mg/kg/d from day +76 to day +87, without further dose adjustments, achieving a progressive resolution of the clinical picture, with normalization of the bilirubin level. On day +89 an ultrasound echotomography showed absence of ascites. Peripheral and bone marrow examinations on day +83 after the infusion of GO, showed a partial complete remission (CRp). The mechanisms for GO-associated VOD are unknown. A direct toxic effect of calicheamicin on the liver has been hypothesized on the basis of the co-expression of CD33 antigen on hepatic sinusoidal endothelial cells, which may be eventually favoured by the use of higher doses of the immunoconjugate (Rajvanshi et al, 2002). Pre-existing liver injury may represent a risk factor for the development of VOD after pretransplant chemotherapeutic conditioning regimens, but it appears to be less important for the occurrence of VOD associated with GO (Giles et al, 2001). The atypical late occurrence of VOD in this patient emphasizes the need for the close monitoring of AML patients treated with GO, even for weeks after the first month from the start of therapy. Defibrotide has been found to be an effective treatment for VOD following bone marrow/stem cell transplantation. The successful outcome after low-dose defibrotide in this patient supports this therapeutic option in the management of VOD associated with GO. Department of Oncology and Haematology, Section of Haematology, University of Modena and Reggio Emilia, 4100 Modena, Italy. E-mail: [email protected]

Alessia Saviola Mario Luppi Leonardo Potenza Monica Morselli Angela Ferrari Giovanni Riva Giuseppe Torelli

753

E.A. (2002) Gemtuzumab ozogamicin monotherapy for relapsed AML after hematopoietic stem cell transplant: efficacy and incidence of hepatic veno-occlusive disease. Bone Marrow Transplantation, 30, 23–28. Giles, J.F., Kantarjian, H.M., Kornblau, S.M., Thomas, D.A., GarciaManero, G., Waddelow, T.A., David, C.L., Phan, A.T., Colburn, D.E., Rashid, A. & Estey, E.H. (2001) Mylotarg therapy is associated with hepatic venoocclusive disease in patients who have not received stem cell transplantation. Cancer, 92, 406–413. Larson, R.A., Boogaerts, M., Estey, E., Karanes, C., Stadmauer, E.A., Sievers, E.L., Mineur, P., Bennett, J.M., Berger, M.S., Eten, C.B., Munteanu, M., Loken, M.R., Van Dongen, J.J., Bernstein, I.D. & Appelbaum, F.R. Mylotarg Study Group (2002a) Antibodytargeted chemotherapy of older patients with acute myeloid leukemia in first relapse using Mylotarg (gemtuzumab ozogamicin). Leukemia, 16, 1627–1636. Larson, R.A., Sievers, E.L., Stadtmauer, E.A., Lowenberg, B., Estey, E., Dombret, H., Theobald, M., Voliotis, D., Leopold, L.H., Richie, M., Brger, M.S., Sherman, M.L. & Appelbaum, F.R. Mylotarg Study Group (2002b) A final analysis of the efficacy and safety of gemtuzumab ozogamicin in 277 patients with acute myeloid leukemia in first relapse. Blood, 100, 1312a. Rajvanshi, P., Shulman, H.M., Sievers, E.L. & McDonald, G.B. (2002) Hepatic sinusoidal obstruction after gemtuzumab ozogamicin therapy. Blood, 99, 2310–2314. Sievers, E.L., Larson, R.A., Stadtmauer, E.A., Estey, E., Lowenberg, B., Dombret, H., Karanes, C., Theobald, M., Bennett, J.M., Sherman, M.L., Berger, M.S., Esten, C.B., Loken, M.R., van Dongen, J.J., Bernstein, I.D. & Appelbaum, F.R. Mylotarg Study Group (2001) Efficacy and safety of gemtuzumab ozogamicin in patients with CD33-positive acute myeloid leukemia in first relapse. Journal of Clinical Oncology, 19, 3244–3254. Wadleigh, M., Richardson, P.G., Zahrieh, D., Lee, S.J., Cutler, C., Ho, V., Alyea, E.P., Antin, J.H., Stone, R.M., Soiffer, R.J. & DeAngelo, D.J. (2003) Prior gemtuzumab ozogamicin exposure significantly increases the risk of veno-occlusive disease in patients who undergo myeloablative allogeneic stem cell transplantation. Blood, 8. Zein, N., Sinha, A.M., McGahren, W.J. & Ellestad, G.A. (1988) Calicheamicin gamma II: an antitumor antibiotic that cleaves double-stranded DNA site specifically. Science, 240, 1198– 1201.

Keywords: veno-occlusive disease, leukaemia, gemtuzumab ozogamicin, defibrotide, Mylotarg.

REFERENCES Cohen, A.D., Luger, S.M., Sickles, C., Mangan, P.A., Porter, D.L., Schuster, S.J., Tsai, D.E., Nasta, S., Gewirtz, A.M. & Stadtmauer,


2003 Blackwell Publishing Ltd, British Journal of Haematology 123: 745–753