Clinical heterogeneity in chronic myeloid leukaemia reflecting biological diversity in normal persons

Clinical heterogeneity in chronic myeloid leukaemia reflecting

biological diversity in normal persons

Myrtle Y. Gordon, Stephen B. Marley, Jane F. Apperley, David Marin, Jaspal Kaeda,

Richard Szydlo and John M. Goldman Leukaemia Research Fund Centre and Department of Haematology, Imperial

College Faculty of Medicine, Hammersmith Campus, London, UK

Received 25 March 2003; accepted for publication 13 April 2003

Summary. The molecular basis of chronic myeloid leukae-mia (CML) is well defined and highly consistent, yet prog-nosis varies considerably. This could reflect the biologicaldiversity occurring in normal populations. We used a col-ony replating assay to measure the proliferative capacity ofprogenitor cells from 211 CML patients and 86 normalpersons. Results were expressed as the frequency distribu-tions of the proliferation index (PI) for individual cases.Normal PI values varied among individuals but werereproducible in individuals. The PIs for CML patients weremoderately but significantly greater (P ¼ 0Æ004) than nor-mal values, consistent with increased progenitor cell pro-liferation in CML. Exposure of CML progenitor cells tothe Abl-kinase inhibitor imatinib shifted their PI towardsthe normal range, implicating p210BCR–ABL. as a cause of

the increased PI. The PIs of CML patients were higher thanthose of their human leucocyte antigen (HLA)-matchedsiblings PI (P ¼ 0Æ003) and patient PI increased exponen-tially with sibling PI (r ¼ 0Æ77; P ¼0Æ001), but not with thePI values of HLA-matched unrelated individuals (P ¼ 0Æ66).Finally, patients with high-risk prognostic scores (accordingto the Sokal or Hasford systems) had a significantly higherPI than those with low risk scores (P ¼ 0Æ01 and 0Æ03respectively). We conclude that heterogeneity in the CMLpatient population is analogous to the constitutionaldiversity in normal subjects.

Keywords: CML, leukaemia, haemopoiesis, diversity,prognosis.

It is a well known, but little understood, fact that patientswith leukaemia vary widely in the biology of their disease andin their responses to therapy. Moreover, heterogeneity alsoexists in many haematological single gene disorders, such asthalassaemia, sickle cell disease and Gaucher’s disease, evenwhen different patients share the same gene mutation(Weatherall, 2000; Beutler, 2001). These observationssuggest that disease heterogeneity might, in some cases,reflect the diversity that exists in the normal population.

Chronic myeloid leukaemia (CML) is a relatively welldifferentiated myeloproliferative disorder originating in atransformed haemopoietic stem cell (Deininger & Goldman,1998). It follows a fairly benign course for several years(chronic phase) before transforming into the more aggres-sive accelerated and blastic phases. The molecular defect inCML is well characterized and quite uniform. It involves the

formation of a fusion gene (BCR-ABL) created by thereciprocal exchange of BCR and ABL gene sequencesbetween chromosomes 9 and 22. The BCR-ABL gene codesfor a protein (p210BCR–ABL) that has heightened proteintyrosine kinase activity. These consistent molecular changesare presumed to be responsible for the pathogenesis of thedisease. In contrast, the biological characteristics of indi-vidual patients are well known to be heterogeneous.Understanding what causes this variation is importantbecause it might influence prognosis and thus lead toimprovements in therapy.

The variable proliferative capacity of myeloid progenitorcells, which is important for expansion of the leukaemic cellpopulation (Gordon et al, 1999), is one facet of thebiological heterogeneity in CML patients. We have devel-oped an assay based on secondary colony formation byindividual progenitor cell-derived clones in semisolid cul-tures that provides the proliferation index (PI) of normalmyeloid progenitor cells (Gordon et al, 1998). As CMLprogenitor cells retain the ability to form colonies insemisolid cultures, the PIs of normal and leukaemicprogenitor cells can be compared directly.

Correspondence: Professor M. Y. Gordon, Leukaemia Research Fund

Centre, Department of Haematology, Imperial College Faculty of

Medicine, Hammersmith Campus, DuCane Road, London W12

0NN, UK. E-mail: [email protected]

British Journal of Haematology, 2003, 122, 424–429

424 � 2003 Blackwell Publishing Ltd

In this study we investigated the hypothesis that differ-ences in progenitor cell PI among CML patients could, atleast in part, reflect the diversity among normal individuals.This normal diversity may be presumed to reflect multiplefactors, both genetic and environmental, that determine thephenotype of a particular individual (Beutler, 2001).Accordingly, we sought to determine whether there weresimilar levels of heterogeneity in the PIs of normal individ-uals and CML patients, and whether there was an influenceof their constitutional background. We also investigatedwhether the PI values of CML patients segregated accordingto the risk groups defined on clinical criteria by Sokal andHasford (Sokal et al, 1984; Hasford et al, 1998).

Overall, the results supported the notion that hetero-geneity among CML patients reflects diversity in thenormal population and that the PI of CML progenitor cellsmay be a biological correlate of the clinical parametersused to construct the Sokal and Hasford prognostic scoringsystems.

MATERIALS AND METHODS

Samples for investigation. Normal bone marrow sampleswere obtained from individuals donating cells for trans-plantation. Peripheral blood samples were obtained fromchronic phase CML patients who were untreated or onlytreated with hydroxyurea, and from treated patients newlypresenting with advanced phase disease in the outpatientclinic of the Department of Haematology, HammersmithHospital. All samples were obtained with informed consent.Information about the BCR-ABL molecular subtype (b2a2versus b3a2 transcript junction), and clinical details toallow calculation of Sokal and Hasford prognostic scores(Sokal et al, 1984; Hasford et al, 1998), was obtained fromclinical records.

Measurement of proliferation index. The method has beendescribed in full elsewhere (Gordon et al, 1998). Primarygranulocyte–-macrophage colony (CFU-GM) cultures wereset up in methylcellulose supplemented with cytokinesusing standard methodology. After 7 d growth, 90 individ-ual colonies were plucked from the methylcellulose andtransferred into fresh methylcellulose and cytokines inseparate wells of 96-well microtitre plates. Each colony wasthoroughly dispersed to single cellularity and the microtitreplates were incubated for a further 7 d. Then, secondarycolonies that had formed from new progenitor cellscontained within the dispersed primary colony were coun-ted. The numbers of secondary colonies per well wereplotted as a cumulative distribution and the area-under-the-curve (AUC), calculated using the trapezium rule, providedthe PI. The PI was determined for each normal and CMLsample. In some experiments, 0Æ1 lmol/l of the p210BCR–ABL

tyrosine kinase inhibitor, imatinib (Druker et al, 1996)(a kind gift from Novartis, Basel, Switzerland) was added toprimary CFU-GM cultures in parallel to cultures withoutimatinib.

Statistical analysis. Data from different groups was com-pared by paired and unpaired two-tailed Student’s t-tests asappropriate. Similarities between groups of donors and

recipients of transplanted stem cells were tested by Spear-man’s rank correlation coefficient. P-values < 0Æ05 wereconsidered to be statistically significant.

RESULTS

Age and sex matchTable I shows the demographic data for 211 chronic phaseCML patients and 86 normal individuals. There was nodifference in the sex ratio between the groups. The normalindividuals were slightly, but significantly, younger thanthe CML patients.

Heterogeneity in normal and chronic phase CML populationsSecondary colony formation by progenitor cells from 86individual normal bone marrow samples and 211 chronicphase (CP) CML samples was evaluated by obtaining the PIfor each case, as described in the Materials and Methods. Theindividual values for each group were then plotted as acumulative frequency distribution (Fig 1). The results show

Table I. Demographics of the CP-CML patients and normal indi-

viduals studied.

CP-CML Normal

Number 211 86

Sex ratio (M/F) 2Æ0 : 1 2Æ3 : 1 P ¼ 0Æ58*

Age (years) 40Æ7 ± 12Æ2� 35Æ7 ± 11Æ2 P ¼ 0Æ006�Median age (years) 39Æ0 32Æ3PI (units)§ 94Æ4 ± 55Æ2� 63Æ9 ± 29Æ9 P ¼ 0Æ004�

*Chi-squared test.

�Student’s t-test.

�Mean ± standard deviation.

§Proliferation index.

Fig 1. Distributions of the magnitude of the proliferation index (PI)

for normal individuals and CML patients with chronic phase and

disease.

Diversity in CML Patients and Normal Persons 425

� 2003 Blackwell Publishing Ltd, British Journal of Haematology 122: 424–429

that the distribution for CP-CML was significantly right-shifted compared with the normal distribution (P ¼ 0Æ003).When cells from individual CP-CML patients were tested ontwo separate occasions, up to 1 month apart, the two PIsmeasured did not differ in magnitude (P ¼ 0Æ4). Also, therewas no difference in the PI distribution for patients whowere untreated and those treated with hydroxyurea(P ¼ 0Æ6). This result was consistent with increased self-renewal by CML progenitor cells above normal values.

The PI of CML progenitor cells is related to the activityof p210 BCR–ABL

To confirm that the level of proliferation by CML progenitorcells is related to the expression of p210BCR–ABL, we testedthe effect of imatinib treatment in vitro on the PI ofprogenitor cells from 99 patients. Imatinib shifted the PIdistribution to the left, thus reducing CML progenitor cellproliferation (P ¼ 0Æ0001, Fig 2). In addition, data from 17patients who were subsequently treated with imatinib for upto 8 weeks showed that the PI of treated patients wassignificantly reduced compared with their pretreatmentvalues (data not shown). In 15 cases, patients’ cells weretested on two separate occasions, up to 1 month apart.There was no difference in sensitivity between the twosamples (P ¼ 0Æ99; paired t-test). Also, there was nodifference in imatinib sensitivity between cells fromuntreated patients and cells from patients treated withhydroxyurea (P ¼ 0Æ2). Thus, inhibition of p210BCR–ABL

kinase activity reduced the proliferative activity of CMLprogenitor cells in vivo and in vitro.

Level of proliferation by CML progenitor cells correlates withproliferation by progenitor cells from their human leucocyteantigen (HLA)-matched compatible siblingsThe data shown in Fig 1 do not reveal whether a CMLpatient with a high PI had a constitutionally high or low PIbefore the development of leukaemia. Since it is reasonable

to expect that members of a family might share certainconstitutional factors to some degree, we compared the PIsof CML patients and their fully HLA-matched histocompat-ible sibling donors. The results in Fig 3A show that there isa significant exponential relationship between the prolifer-ative capacity of progenitor cells from CP-CML patients andtheir siblings (P ¼ 0Æ001), and that the patient’s PI washigher than that of the normal sibling in the great majorityof cases (P ¼ 0Æ003). This result suggested that a CMLpatient with a high PI may have had a high normal PI priorto developing the disease. In contrast, no relationshipexisted between the PIs of CML patients and those of theirHLA-matched histocompatible, but unrelated, donors(Fig 3B). As these results indicate that histocampatibilityantigens do not determine variation in the PI, we randomlyallocated ‘donors’ from the group of normal individuals tothe CP-CML patients, using random number generation,and investigated the correlation between their PI values.The data in Fig 3C are the result of 100 iterations and showthe distribution of the correlations obtained. Accordingly,the probability of obtaining a significant correlation was4%. Similarly, we randomly paired normal donors with oneanother and obtained a significant association in 2% of theiterations (Fig 3D). These observations support the idea thatconstitutional factors influence the biological phenotype ofCML.

Relationship of PI to prognostic scoresAs CML is a disease of progenitor cell proliferation weinvestigated whether the PI values had any prognosticsignificance. The data in Fig 4A and B show that progenitorcell proliferation was significantly greater in risk group 3than in risk group 1, with intermediate levels in risk group2, both for the Sokal and for the Hasford scoring systems.Thus, the PI of progenitor cells in CML may be the biologicalcorrelate of the clinical parameters used to predict progno-sis. It has been suggested that the precise position of thebreakpoint in BCR resulting in a b2a2 or b3a2 fusiontranscript has prognostic significance, but there was nodifference in PI between these two groups of patients (b2a2,n ¼20; b3a2, n ¼29; P ¼0Æ57).

Is a high PI a risk factor for the development of CML?The correspondence between the PI and prognostic riskgroups (Fig 4) and the relationship between the PIs of CMLpatients and their HLA-matched siblings raise the possibilitythat a high PI in normal individuals could be a risk factor forthe development of CML. This question is difficult to address,but we found no significant difference when we comparedthe PIs of 39 donors who had an HLA-matched sibling withCML to the PIs of 37 donors who did not (P ¼ 0Æ65).

DISCUSSION

In this paper we have explored the idea that intrinsicdiversity in normal human populations may have an impacton the widely observed heterogeneity in disease phenotypes.This diversity is likely to result from combinations ofmultiple factors, which themselves may vary in different

Fig 2. Influence of imatinib treatment (imatinib-tx) on the magni-

tude of the PI of chronic phase CML patients.

426 M. Y. Gordon et al


people. Chronic myeloid leukaemia is an ideal condition forstudy in this regard because it fulfils the necessary require-ments, notably access to large numbers of patients andnormal individuals and an endpoint that can be studied inboth groups. During the chronic phase of CML, myeloidprogenitor cells retain their ability to form colonies insemisolid medium in vitro, so that it is possible to measurethe PI of colony-forming cells from normal individuals andCML patients in the same way. It is relevant that an increasein PI would be predicted to result in the myeloid expansionthat is characteristic of CML (Gordon et al, 1998, 1999).

Our first comparison revealed that the PI of progenitorcells was highly variable in both CML patients and normalindividuals, but was greater in the CML group than thenormal group. This observation suggests that the increasedPI is a consequence of the expression of p210BCR–ABL proteintyrosine kinase. Imatinib (formerly known as STI-571) is aninhibitor of ABL tyrosine kinases which specifically inhibitsp210BCR–ABL-positive cells while sparing normal cells (Dru-ker et al, 1996; Marley et al, 2000). Imatinib treatmentshifted the CML PI distribution curve to the left, whichconfirms its action on the leukaemic progenitors and

Fig 3. (A) Relationship between the magnitude of the PI of chronic phase CML patients and the PI of their HLA-matched, non-leukaemic

siblings. The data best fitted an exponential relationship (solid line). The CML patients PI values were greater than those of their siblings in the

great majority of cases, as shown by the broken line that indicates equal PI values for patients and their siblings. (B) Lack of relationship

between the PI of chronic phase CML patients and their HLA-matched unrelated donors. (C) Distribution of correlations between PI values

obtained by randomly pairing data from CML patients with data from normal individuals. Significant correlations were obtained in 4 (above

and below the horizontal lines) of 100 iterations (simulations). (D) Distribution of correlations between PI values obtained by randomly pairing

data from normal individuals one another. Significant correlations were obtained in 2 (above and below the horizontal lines) of 100 iterations

(simulations).



supports the view that increased proliferation in CML isattributable to p210BCR–ABL.

Although the PI in CML was greater than that of normalprogenitor cells, there was considerable overlap in the PImagnitudes of normal individuals’ and patients’ cells. Oneinterpretation of this finding is that only a subset of CMLpatients is distinct from the normal donors. Alternatively,individual patients’ PI values could reflect their PI prior todeveloping CML. The distributions shown in Fig 1 did notprovide any information about the PI of an individualpatient’s cells prior to the development of leukaemia. Wereasoned that constellations of constitutional factors couldbe relatively restricted to individuals and their closerelatives. There was a significant exponential relationshipbetween the PI values measured in CML patients and thoseof their fully HLA-matched sibling donors, although the PIsof the siblings were significantly lower (P ¼ 0Æ003) thanthose of the patients. While normal bone marrow isconsidered the appropriate control for CML blood, it is

noteworthy that the PI of progenitor cells in normal blood isconsiderably lower than the PI of progenitor cells in bonemarrow (Gordon et al, 1998). We also discriminatedbetween patients with b2a2 and b3a2 fusion junctionsbut did not identify separate groups with respect to themagnitude of the PI. These results suggest that the kineticproperties of a CML patient’s leukaemic progenitor cells mayreflect, to a degree, the kinetic properties of the normalprogenitor cell population in which the 9; 22 chromosomaltranslocation occurred.

The fact that imatinib treatment returned the PI to thenormal distribution and that the ratio between the patients’and their donors’ PI was moderate [1Æ8 ± 0Æ25;mean + standard error of the mean (sem)] suggests thatthe action of p210BCR–ABL approximately doubles theproliferative capacity of progenitor cells in CML. Thisrelatively small change is consistent with CP-CML being achronic disease and probably associated with quite subtlechanges in haemopoietic cell kinetics, and with the uniformmolecular characteristics of the disease. However, prelim-inary data indicated a further increase in PI in 17 patientswith accelerated or blast phase disease (P ¼ 0Æ6; data notshown).

Overall, the data suggest that variation among normalindividuals may contribute to inter patient heterogeneity inCML. There is much evidence from murine studies thathaemopoietic progenitor cells in different mouse strainsvary widely in number and proliferative activity (de Haanet al, 2000; Morrison et al, 2002). Therefore, it is notsurprising to find extensive variation among normalhumans. Koller et al (1996) found that cell production byhuman stem cells in vitro varied by two orders of magnitude,which is similar to the degree of variation shown by ourdata. One reason why we and Koller et al (1996) were ableto identify the range of normal variation may lie in the largenumbers of individuals studied: we investigated 86 normaldonors while Koller et al (1996) investigated 52. Manystudies on human progenitor cells are based on muchsmaller numbers, which would not reveal the true extent ofnormal variation.

Differences in behaviour of haemopoietic progenitor cellsfrom different normal individuals may be attributable togenetic diversity or other variables. Koller et al (1996) wereable to eliminate sex, age, weight and height as contributingvariables. We have previously shown that the PI of humansfalls as a function of age (Marley et al, 1999). However, therestricted age range of patients and donors in this studymeans that age is unlikely to contribute greatly to theobserved variation. Also, the PI of normal males was notsignificantly greater than that of normal females (data notshown), which is consistent with the findings of Koller et al(1996). In addition, the observation that the PI of CMLpatients does not correlate with that of HLA-matchedunrelated donors but does correlate with that of their HLA-matched siblings argues that histocompatibility antigensmay also be excluded.

More detailed studies on different strains of mice havereported associations between genetic markers and thefrequency and activity of haemopoietic stem and progenitor

Fig 4. Segregation of the magnitude of the PI of chronic phase CML

patients according to the risk groups defined by the Sokal (A) and

Hasford (B) prognostic scoring systems.

428 M. Y. Gordon et al


cells. Recently, de Haan et al (2002) concluded that theexpression levels of a large number of genes might beresponsible for controlling stem cell behaviour. Thesecollections of genes may be analogous to those responsiblefor the interindividual diversity in progenitor cell behaviourseen by Koller et al (1996) and ourselves in humans.

Heterogeneity among patients is the raw material ofprognostic scoring systems because if it did not exist allpatients would respond in the same way and prognosticscores would be unnecessary. We found that the magnitudeof the PI in individual CML patients corresponded to theirprognostic risk group as defined by the Sokal and Hasfordprognostic scoring systems (Sokal et al, 1984; Hasford et al,1998). Therefore, the PI in CML may be a biologicalcorrelate of the clinical parameters currently used forprognosis. While these associations might be taken to implythat a high PI is a risk factor for the development of CML,there was no difference in the PI distributions of normalindividuals who had an HLA-identical sibling with CML andthose who did not. A number of studies have examined thehypothesis that the type of BCR-ABL fusion transcript (b2a2versus b3a2) influences the duration or survival of chronicphase disease and response to treatment with interferonalpha (Dowding et al, 1991; Shepherd et al, 1995; Millset al, 1998), with conflicting results, but the present studydoes not suggest that the type of fusion transcript influencesbiological properties of CML progenitor cells.

In summary, normal individuals exhibit considerablediversity, which may be reflected by the apparent hetero-geneity of diseases such as CML. The factors responsible fordetermining the diversity of individuals are likely to be verydifficult to identify but they should be considered indiscussions about disease heterogeneity.

ACKNOWLEDGMENTS

The work was supported by a Specialist Programme Grantfrom the Leukaemia Research Fund.

REFERENCES

Beutler, E. (2001) Discrepancies between genotype and phenotype

in hematology: an important frontier. Blood, 98, 2597–2602.

Deininger, M.W.N. & Goldman, J.M. (1998) Chronic myeloid leu-

kemia. Current Opinion in Hematology, 5, 302–308.

Dowding, C., Guo, A.-P., Maisin, D., Gordon, M.Y. & Goldman, J.M.

(1991) The effects of interferon-a on the proliferation of CML

progenitor cell in vitro are not affected by the precise position of

the M-BCR breakpoint. British Journal of Haematology, 77, 165–

171.

Druker, B.J., Tamura, S., Buchdunger, E., Ohno, S., Segal, G.M.,

Fanning, S., Zimmerman, J. & Lydon, N.B. (1996) Effects of a

selective inhibitor of the abl tyrosine kinase on the growth of bcr-

abl positive cells. Nature Medicine, 2, 561–566.

Gordon, M.Y., Marley, S.B., Lewis, J.L., Davidson, R.J., Nguyen,

D.X., Grand, F.H., Amos, T.A.S. & Goldman, J.M. (1998) Treat-

ment with interferon-a preferentially reduces the capacity for

amplification of granulocyte-macrophage progenitors (CFU-GM)

from patients with chronic myeloid leukemia but spares normal

CFU-GM. Journal of Clinical Investigation, 102, 710–715.

Gordon, M.Y., Dazzi, F., Marley, S.B., Lewis, J.L., Nguyen, D.X.,

Grand, F.H., Davidson, R.J. & Goldman, J.M. (1999) Cell biology

of CML cells. Leukemia, 13, 65–71.

de Haan, G., Szilvassy, S.J., Meyerrose, T.E., Dontje, B., Grimes, B. &

Van Zant, G. (2000) Distinct functional properties of highly

purified hematopoietic stem cells from mouse strains differing in

stem cell numbers. Blood, 96, 1374–1379.

de Haan, G., Bystrykh, L.V., Weersing, E., Dontje, B., Geiger, H.,

Ivanova, N., Lemischka, I., Vellenga, E. & Van Zant, G. (2002) A

genetic and genomic analysis identifies a cluster of genes asso-

ciated with hematopoietic cell turnover. Blood, 100, 2056–2062.

Hasford, J., Pfirrman, M., Hehlmann, R., Allen, N.C., Baccarani, M.,

Kluin-Nelemans, J.C., Alimena, G., Steegman, J.L. & Ansari, H.

(1998) A new prognostic score for survival of patients with

chronic myeloid leukemia treated with interferon alfa. Journal of

the National Cancer Institute, 90, 850–858.

Koller, M.R., Manchel, I., Brott, D.A. & Palsson, B.O. (1996) Donor-

to-donor variability in the expansion potential of human bone

marrow cells is reduced by accessory cells but not by soluble

growth factors. Experimental Hematology, 24, 1484–1493.

Marley, S.B., Lewis, J.L., Davidson, R.J., Roberts, I.A.G., Dokal, I. &

Goldman, J.M. (1999) Evidence for a continuous decline in

haemopoietic cell function from birth: application to evaluating

bone marrow failure in children. British Journal of Haematology,

106, 162–166.

Marley, S.B., Deininger, M.W.N., Davidson, R.J., Goldman, J.M. &

Gordon, M.Y. (2000) The tyrosine kinase inhibitor STI571, like

interferon-a, preferentially reduces the capacity for amplification

of granulocyte-macrophage progenitors from patients with

chronic myeloid leukemia. Experimental Hematology, 28, 551–

557.

Mills, K.I., MacKenzie, E.D. & Birnie, G.D. (1998) The site of the

breakpoint within bcr is a prognostic factor in Philadelphia-

positive CML patients. Blood, 72, 1237–1241.

Morrison, S.J., Quian, D., Jerabek, L., Thiel, B.A., Park, I.-K., Ford,

P.S., Kiel, M.J., Schork, N.J., Weissman, I.L. & Clarke, M.F.

(2002) A genetic determinant that specifically regulates the

frequency of hematopoietic stem cells. Journal of Immunology,

168, 635–642.

Shepherd, P., Suffolk, R., Halsey, J. & Allan, N. (1995) Analysis

of molecular breakpoint and m-RNA transcripts in a pro-

spective randomised trial in chronic myeloid leukemia: no

correlation with clinical features, cytogenetic response, dura-

tion of chronic phase or survival. British Journal of Haematol-

ogy, 89, 546–554.

Sokal, J.E., Cox, E.B., Baccarani, M., Tura, S., Gomez, G.A.,

Robertson, J.E., Tso, C.Y., Braun, T.J., Clarkson, B.D. & Cervantes,

F. (1984) Prognostic discrimination in ‘good risk’ chronic gran-

ulocytic leukemia. Blood, 63, 789–799.

Weatherall, D.J. (2000) Single gene disorders or complex traits:

lessons from the thalassemias and other monogenic diseases.

British Medical Journal, 321, 1117–1120.



Documents

Clinical heterogeneity in chronic myeloid leukaemia reflecting biological diversity in normal persons