Acute myeloid leukaemia (AML) varies greatly in response to therapy. Although cytogenetics is the most important predictor of response, it is least informative in the 40 - 50% of adults with normal cytogenetics (Grimwade et al, 1998). However, an increasing number of molecular markers with prognostic significance, particularly in cytogenetically normal AML (CN-AML), have been identified. The two most common are internal tandem duplications (ITD) in the FLT3 gene (FLT3-ITD) and mutations in the nucleophosmin gene (NPM1) with frequencies of approximately 50% and 30%, respectively, in CN-AML. While FLT3-ITD is associated with short relapse-free and overall survival, the prognosis of CN-AML with NPM1 mutations without FLT3-ITD (NPM+/FLT3-ITD-) is similar to that of “favourable” cytogenetics AML (Kottaridis et al, 2001;Schnittger et al, 2005; Thiede et al 2006).
NPM1 and FLT3-ITD are often used to guide decisions about therapy in first complete remission (CR). In particular, the high relapse rate of FLT3-ITD+ AML leads to use of allogeneic haematopoietic cell transplant (HCT) while the lower relapse rate of NPM+/FLT3-ITD- leads to administration of repetitive cycles of high dose cytarabine (Brunet et al, 2012; Döhner et al, 2010).
Despite their prognostic relevance, FLT3-ITD and NPM1 status at diagnosis is often unknown when patients are referred in first CR to tertiary centres. Here we report efforts to develop a multivariate model based on other clinical and laboratory data to predict probability of a patient being FLT3-ITD+ or NPM1+/FLT3-ITD-. Our hypothesis was that examining covariates in combination would add to predictive ability in contrast to examining them singly.
Our analysis included 434 adults with newly diagnosed AML (excluding acute promyelocytic leukaemia) treated on four Southwest Oncology Group (SWOG) trials (S9031, S9333, S9500, and S0106). Details of the treatment protocols have been previously reported (Godwin et al, 1998; Anderson et al, 2002; Petersdorf et al, 2007; Petersdorf et al, 2009). Two independent validation cohorts were 1,401 patients treated on various protocols at the MD Anderson Cancer Center (MDA) and 210 patients treated on various protocols at Fred Hutchinson Cancer Research Center (FHCRC). Institutional review boards of participating institutions approved all protocols, and patients were treated according to the Declaration of Helsinki. FLT3-ITD and NPM1 status (positive/negative) was established at diagnosis in all patients included in this study.
Univariate and multivariate logistic regression models were used to identify covariates associated with being (1) FLT3-ITD+ and (2) NPM1+/FLT3-ITD- at diagnosis. Covariates examined were age, sex, performance status (PS), white blood cell count (WBC), platelet count, bone marrow blast percentage, secondary AML, and cytogenetic risk (using SWOG criteria). The relative ability of covariates to predict outcomes was quantified using the area under the receiver operator characteristic curve (AUC); an AUC of 1.0 denotes perfect prediction, whereas an AUC of 0.5 is analogous to a coin flip. The scoring systems were tested in the two independent populations (MDA and FHCRC).
Characteristics of the 428 patients treated on SWOG trials are summarized in Table I. FLT3-ITD was present in 23.6% (101 out of 428). WBC>20 × 109/l and normal cytogenetics had the largest AUCs for predicting FLT3-ITD positivity: 0.65 and 0.69, respectively (Table I). A multivariate model indicated that WBC > 20 × 109/l (odds ratio 3.4 vs WBC < 20 × 109/l, p <0.001), SWOG favourable cytogenetics (odds ratio 0.06 vs normal cytogenetics, p <0.001) and SWOG unfavourable cytogenetics, (odds ratios 0.23 vs normal cytogenetics, p < 0.001) were independently associated with being FLT3-ITD+ at diagnosis. The AUC of a model combining these features was 0.79 and contrasted with 0.65 and 0.69, respectively, when only WBC or cytogenetics were considered.
Table 1. Covariates associated with being FLT3-ITD positive.
| Variables | FLT3-ITD- N=327(%) | FLT3-ITD+ N=101 (%) | p-value | AUC |
|---|---|---|---|---|
| Age (years) | ||||
| ≤50 | 155 (47) | 47 (47) | 0.91 | 0.5 |
| >50 | 172 (53) | 54 (53) | ||
| Sex | ||||
| Female | 154 (47) | 48 (48) | 1 | 0.5 |
| Male | 173 (53) | 53 (52) | ||
| ECOG Score | ||||
| 0 | 105 (32) | 28 (28) | 0.15 | 0.55 |
| 1 | 172 (53) | 49 (49) | ||
| 2- 3 | 50 (15) | 24 (24) | ||
| Cytogenetics | ||||
| Normal | 128 (39) | 71 (70) | <0.001 | 0.69 |
| Favourable | 56 (17) | 2 (2) | ||
| Intermediate (non-normal) | 22 (7) | 10 (10) | ||
| Unfavourable | 95 (29) | 7 (7) | ||
| Unknown | 26 (8) | 11 (11) | ||
| Platelet count | ||||
| ≤50 × 109/l | 159 (49) | 44 (44) | 0.43 | 0.53 |
| >50 × 109/l | 168 (51) | 57 (56) | ||
| WBC | ||||
| ≤20 × 109/l | 194 (59) | 29 (29) | <0.001 | 0.65 |
| >20 × 109/l | 133 (41) | 72 (71) | ||
| BM Blast (%) | ||||
| ≤70 | 192 (59) | 43 (43) | 0.006 | 0.58 |
| >70 | 135 (41) | 58 (57) | ||
| De novo AML | 301 (92) | 98 (97) | 0.11 | 0.52 |
| Secondary AML | 26 (8) | 3 (3) |
FLT-ITD, FLT-internal tandem duplication; AUC, area under the receiver operator characteristic curve; ECOG, Eastern Cooperative Oncology Group; WBC, white blood cell count; BM, bone marrow; AML, acute myeloid leukaemia
We then tested the multivariate model's performance in the MDA and FHCRC populations that had not contributed to the system's development and in whom FLT3-ITD status at diagnosis was known. In the FHCRC population (16% FLT3-ITD positive) its AUC was only 0.58, not better than when each covariate was examined separately (AUC 0.54 and 0.6 for WBC and cytogenetics, respectively). Similarly, at MDA (17% FLT3-ITD positive), the system's AUC was 0.68 versus AUC of 0.59 and 0.68 for WBC and cytogenetics, respectively.
A similar approach was used to identify covariates associated with being NPM1+/FLT3-ITD-. Characteristics of SWOG patients with NPM1+/FLT3-ITD- and FLT3-ITD+ (with or without NPM1) or FLT3-ITD-/NPM1- treated on SWOG trials are summarized in Table II. NPM1+/FLT3-ITD- was present in 16% (63 of the 401). Cytogenetics was associated with NPM1+/FLT3-ITD- with AUC of 0.73 and multivariate analysis indicated that the only covariate independently associated with NPM1+/FLT3-ITD- was cytogenetics. As the AUC of the multivariate model (0.79) was only slightly better than what had been observed with cytogenetics alone, we did not attempt to verify the model in the MDA and FHCRC populations.
Table II. Covariates associated with being NPM1 positive and FLT3-ITD negative.
| Variables | FLT3-ITD and NPM1 known but not NPM+/FLT3-ITD- N=338 (84%) | NPM1+/FLT3-ITD-N=63 (16%) | p-value | AUC |
|---|---|---|---|---|
| Age (years) | ||||
| ≤50 | 147 (43) | 27 (43) | 1 | 0.5 |
| >50 | 191 (57) | 36 (57) | ||
| Sex | ||||
| Female | 156 (46) | 36 (57) | 0.13 | 0.55 |
| Male | 182 (54) | 27 (43) | ||
| ECOG Score | ||||
| 0 | 99 (29) | 21 (33) | 0.24 | 0.55 |
| 1 | 171 (51) | 35 (56) | ||
| 2- 3 | 68 (20) | 7 (11) | ||
| Cytogenetics | ||||
| Favourable | 131 (39) | 50 (79) | <0.001 | 0.73 |
| Intermediate | 57 (17) | 1 (2) | ||
| Unfavourable | 53 (16) | 9 (14) | ||
| Unknown | 97 (29) | 3 (5) | ||
| Platelet count | ||||
| ≤50 × 109/l | 149 (44) | 34 (54) | 0.099 | 0.56 |
| >50 × 109/l | 169 (50) | 24 (38) | ||
| WBC | ||||
| ≤20 × 109/l | 189 (56) | 29 (46) | 0.17 | 0.55 |
| >20 × 109/l | 149 (44) | 34 (54) | ||
| BM Blast (%) | ||||
| ≤70 | 187 (55) | 33 (52) | 0.68 | 0.51 |
| >70 | 151 (45) | 30 (48) | ||
| De novo AML | ||||
| Secondary | 316 (93) | 61 (97) | 0.4 | 0.52 |
| AML | 22 (7) | 2 (3) |
Legend: NPM1, nucleophosmin gene; FLT-ITD, FLT-internal tandem duplication; AUC, area under the receiver operator characteristic curve; ECOG, Eastern Cooperative Oncology Group; WBC, white blood cell count; BM, bone marrow; AML, acute myeloid leukaemia
FLT3-ITD and NPM1 are used to assign patients to HCT (FLT3-ITD+) or repetitive cycles of chemotherapy (NPM1+/FLT3-ITD-). However FLT3-ITD+ and NPM1+ status at diagnosis is often unknown when patients are referred to tertiary centres after reaching CR. This prompted an attempt to develop multitvariate models to predict the probability of a patient being (a) FLT3-ITD+ or (b) NPM1+/FLT3-ITD- based on readily available covariates. Our hypothesis was that a combination of covariates would add to the ability to predict (a) or (b) beyond what could be predicted knowing single covariates, e.g. higher WBC or normal cytogenetics, known to be associated with NPM1+ and FLT3-ITD status.
Using AUC to quantify predictive ability, we found that a combination of WBC > 20 × 109/l and normal cytogenetics was more effective in predicting FLT3-ITD+ in SWOG patients than either WBC or cytogenetics. However this is not true when the model was applied to MDA or FHCRC patients. One possible explanation is that the proportion of FLT3-ITD patients was less at FHCRC (p=0.04) or at MDA (p=0.002) than in SWOG. We were not successful in bettering the predictive ability of cytogenetics for NPM1+/FLT3-ITD-. Hence our hypothesis was not supported, suggesting that there is no obvious substitute for actual data on FLT3-ITD and NPM1.
Footnotes
Authorship: EE designed the research study; MO and EE analysed the data; RS collected the data; HK, FR, PH, SF, PB, JC, JP, SP, JG, CW, SP, AL, RW, DS, FA contributed patients to the studies; MO, EE and FO interpreted data; FO and EE wrote the manuscript; SM, FA, RW and DS revised it critically. All authors approved the submitted and final version.
Conflict-of interest disclosure: The authors declare no competing financial interests.
References
- Anderson JE, Kopecky KJ, Willman CL, Head D, O'Donnell MR, Luthardt FW, Norwood TH, Chen IM, Balcerzak SP, Johnson DB, Appelbaum FR. Outcome after induction chemotherapy for older patients with acute myeloid leukemia is not improved with mitoxantrone and etoposide compared to cytarabine and daunorubicin: a Southwest Oncology Group study. Blood. 2002;100:3869–3876. doi: 10.1182/blood-2001-12-0354. [DOI] [PubMed] [Google Scholar]
- Brunet S, Labopin M, Esteve J, Cornelissen J, Socié G, Iori AP, Verdonck LF, Volin L, Gratwohl A, Sierra J, Mohty M, Rocha V. Impact of FLT3-ITD on the outcome of related and unrelated hematopoietic transplantation for adult AML in first remission: a retrospective analysis. Journal of Clinical Oncology. 2012;30:735–41. doi: 10.1200/JCO.2011.36.9868. [DOI] [PubMed] [Google Scholar]
- Döhner H, Estey EH, Amadori S, Appelbaum FR, Büchner T, Burnett AK, Dombret H, Fenaux P, Grimwade D, Larson RA, Lo-Coco F, Naoe T, Niederwieser D, Ossenkoppele GJ, Sanz MA, Sierra J, Tallman MS, Löwenberg B, Bloomfield CD European LeukemiaNet. Diagnosis and management of acute myeloid leukemia in adults: recommendations from an international expert panel, on behalf of the European LeukemiaNet. Blood. 2010;115:453–474. doi: 10.1182/blood-2009-07-235358. [DOI] [PubMed] [Google Scholar]
- Godwin JE, Kopecky KJ, Head DR, Willman CL, Leith CP, Hynes HE, Balcerzak SP, Appelbaum FR. A double-blind placebo-controlled trial of granulocyte colony-stimulating factor in elderly patients with previously untreated acute myeloid leukemia: a Southwest oncology group study (9031) Blood. 1998;91:3607–15. [PubMed] [Google Scholar]
- Grimwade D, Walker H, Oliver F, Wheatley K, Harrison C, Harrison G, Rees J, Hann I, Stevens R, Burnett A, Goldstone A. The importance of diagnostic cytogenetics on outcome in AML: analysis of 1,612 patients entered into the MRC AML 10 trial. The Medical Research Council Adult and Children's Leukaemia Working Parties. Blood. 1998;92:2322–33. [PubMed] [Google Scholar]
- Kottaridis PD, Gale RE, Frew ME, Harrison G, Langabeer SE, Belton AA, Walker H, Wheatley K, Bowen DT, Burnett AK, Goldstone AH, Linch DC. The presence of a FLT3 internal tandem duplication in patients with acute myeloid leukemia (AML) adds important prognostic information to cytogenetic risk group and response to the first cycle of chemotherapy: analysis of 854 patients from the United Kingdom Medical Research Council AML 10 and 12 trials. Blood. 2001;98:1752–9. doi: 10.1182/blood.v98.6.1752. [DOI] [PubMed] [Google Scholar]
- Petersdorf SH, Rankin C, Head DR, Terebelo HR, Willman CL, Balcerzak SP, Karnad AB, Dakhil SR, Appelbaum FR. Phase II evaluation of an intensified induction therapy with standard daunomycin and cytarabine followed by high dose cytarabine for adults with previously untreated acute myeloid leukemia: a Southwest Oncology Group study (SWOG-9500) American Journal of Hematolology. 2007;82:1056–62. doi: 10.1002/ajh.20994. [DOI] [PubMed] [Google Scholar]
- Petersdorf SH, Kopecky KJ, Slovak M, Willman C, Nevill T, Brandwein J, Larson RA, Erba HP, Stiff PJ, Stuart RK, Walter RB, Tallman MS, Stenke L, Appelbaum FR. A phase III study of gemtuzumab ozogamicin during induction and post-consolidation therapy in younger patients with acute myeloid leukemia. Blood. 2013 Apr 16; doi: 10.1182/blood-2013-01-466706. Epub ahead of print. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Schnittger S, Schoch C, Kern W, Mecucci C, Tschulik C, Martelli MF, Haferlach T, Hiddemann W, Falini B. Nucleophosmin gene mutations are predictors of favorable prognosis in acute myelogenous leukemia with a normal karyotype. Blood. 2005;106:3733–9. doi: 10.1182/blood-2005-06-2248. [DOI] [PubMed] [Google Scholar]
- Thiede C, Steudel C, Mohr B, Schaich M, Schäkel U, Platzbecker U, Wermke M, Bornhäuser M, Ritter M, Neubauer A, Ehninger G, Illmer T. Analysis of FLT3-activating mutations in 979 patients with AML: association with FAB subtypes and identification of subgroups with poor prognosis. Blood. 2002;99:4326–35. doi: 10.1182/blood.v99.12.4326. [DOI] [PubMed] [Google Scholar]