Skip Navigation U.S. Department of Health and Human Services
Agency for Healthcare Research Quality
Archive print banner
Report on the Relative Efficacy of Oral Cancer Therapy for Medicare Beneficiaries Versus Currently Covered Therapy

This information is for reference purposes only. It was current when produced and may now be outdated. Archive material is no longer maintained, and some links may not work. Persons with disabilities having difficulty accessing this information should contact us at: Let us know the nature of the problem, the Web address of what you want, and your contact information.

Please go to for current information.

Part 3. Imatinib for Chronic Myeloid Leukemia (CML) (continued)


The search strategy yielded 418 articles. The selection process is described below:

Identified by search strategy
   |------ Excluded based on review of abstract
   |     (N=162)
Included based on review of abstract
   |------ Unable to locate
   |     (N=8)
   |------ Excluded based on full-text review
   |     (N=89)
   |          23   not phase II-III for efficacy
   |          11   case series not selected on response
   |          2   case series selected on adverse events
   |          25   no quantification of association
   |          5   wrong drug
   |          9   wrong outcomes
   |          2   wrong disease
   |          5   review articles
   |          3   no data reported
   |          4   abstracts superseded by published article
Included in full-text review and evidence tables

The 158 included reports comprised 69 full reports and abstract-only publications cited in Tables 1a-1d, as well as 36 full reports cited in the text of this report. Study designs included one published phase III controlled clinical trial with five sub-studies. The exact number of unique phase II uncontrolled clinical trials is difficult to establish, as many authors presented data from the same groups of subjects in multiple reports. By best assessment there are approximately 30 individual phase II trials presented here. All of the adverse events data were derived from the phase II and III clinical trials that were published in full reports, with the exception of four additional individual adverse event reports (two full-text articles and two abstracts).

Quality of the studies varied by outcome category (Tables 1a-1d, and Appendix B). The main imatinib efficacy studies published in full were of high quality. Quality, in general, was lower for predictor studies, consistent with these more commonly being written as basic science reports with minor clinical correlations. The other group of lower quality reports was emerging reports, especially those evaluating imatinib after stem cell transplant and in the heavily treated setting.


Evidence of efficacy of imatinib in CML can be best considered in terms of the matrix presented in Figure 5. Discrete studies are available for the first three CP clinical settings with the imatinib resistant setting addressed in the Future Directions section. The AP and BP clinical settings are presented in studies of mixed populations, represented on the respective tables. In addition, Table 7 presents those studies of mixed phases and Table 8 includes studies of imatinib combined with other treatments. Table 9 presents efficacy in terms of quality of life.

Chronic phase

The most convincing and highest quality data for imatinib in CML is derived from the large phase III trial of imatinib vs. interferon plus cytarabine published in 2003 by O'Brien, et al., the International Randomized Study of Interferon versus STI571 (IRIS).59 Prior to imatinib, interferon plus cytarabine was considered the standard of care for newly diagnosed CP CML when stem cell transplantation was not possible. In the RCT by Guilhot, et al., published in 1997, interferon plus cytarabine was superior to interferon alone, with 41 percent achieving Major CR for the combined intervention vs. 24 percent for interferon alone (p=0.001), and 15 percent vs. 9 percent Complete CR rates.42 The estimated 3-year overall survival (OS) was 86 percent. The superior intervention from the Guilhot study was then the comparator for the IRIS trial.

In the IRIS phase III trial of imatinib vs. interferon plus cytarabine, imatinib was clearly superior with an 85 percent Major CR rate compared to 22% for interferon plus cytarabine, and Complete CR rates of 74 percent vs. 9 percent. While the OS presented in the original report was not different, PFS at 18 months was significantly better with imatinib (92 percent vs. 74 percent, p<0.0001). In a followup abstract report, the 30-month OS for imatinib was 95 percent (93-97 percent); OS for interferon plus cytarabine was not presented.62 The efficacy of the interferon plus cytarabine arm was not as good in IRIS as in the original Guilhot study (Major CR rate in IRIS 22 percent, Guilhot et al 41 percent), however the 30-month OS rates with imatinib in IRIS (95 percent) are still substantially higher than the 36-month OS rates with interferon and cytarabine from the Guilhot et al RCT (86 percent).

In addition to clinical response rates, other important treatment-related insights that can be derived from this group of studies includes the molecular impact of imatinib, timing of maximal treatment effect, dosing parameters, and tolerability of stem cell transplantation after imatinib.

Imatinib is a targeted drug that interacts with the BCR-ABL tyrosine kinase protein and ultimately leads to apoptosis and destruction of CML cells. Reduction in CML cells should lead to fewer cells with the Ph and therefore fewer cells producing the BCR-ABL mRNA. Reduction in the number of CML cells with Ph is the "cytogenetic response" described previously. Reduction in the number of cells producing the mRNA transcripts is the "molecular response." Evidence of molecular response has been linked to survival,6,11 In terms of imatinib efficacy, Hughes, et al. demonstrated that among IRIS patients who achieved a Complete CR at 6 months, imatinib led to more molecular responses (42 percent vs. 13 percent, p=0.01). Branford and colleagues demonstrated that 71 percent of IRIS participants obtained a Major MR (defined by this group as >3 log reduction in BCR-ABL/ABL transcript numbers) by 3 years. The rate of Major MR continued to increase over the first 2 years, and after that did not appear to increase or decrease substantially.32

Karntarjian, Cortes, and colleagues provided further insight into the timing of the response to imatinib in one of their phase II trial reports.63 Complete CR rates for imatinib increased from 34 percent to 60 percent over the period from 3 to 9 months. Meanwhile, the Complete CR rates for a group of historical controls that received interferon alone did not increase substantially after 3 months and, similarly, little increase was noted for interferon plus cytarabine patients after 6 months. A steady increase in the number of Complete CRs over 12 months was noted within studies of imatinib in the CP interferon refractory setting.8,70,73 Taken together with the molecular response data from IRIS,32 these data suggest that maximal responses to imatinib take longer than was previously seen with interferon-based regimens and that efficacy analyses need to be clearly presented in the context of duration of exposure to imatinib.

Uncontrolled phase II studies support the conclusion of the IRIS trial and provide additional clinical insight into the appropriate starting dose. Most convincingly, Kantarjian, Talpaz and colleagues treated patients with 800 mg of imatinib, and achieved better Complete CR rates as compared to historical controls who had received 400 mg of imatinib (90 percent vs. 74 percent).12 In an ongoing phase II trial by Hughes, et al., the 600 mg dose appears to be leading to higher Major CR and Complete CR rates, as compared to a historical control group from the IRIS trial that received 400 mg.65 Direct comparisons between doses are not currently available.

Finally, possibility of stem cell transplantation after progression on imatinib was evaluated in an abstract presentation by Guilhot, et al.62 Seventy-five IRIS participants went on to stem cell transplantation. There were no differences in survival after transplant between participants who received imatinib (N=30) and those who received interferon with cytarabine (N=45).

Chronic Phase — Interferon resistant or refractory

CML that has been previously treated is expected to be more resistant to the next therapy. Leukemic cells develop genetic or other changes that protect the cell and help them evade subsequent treatments. Hence, imatinib treatment of CML in the interferon resistant or refractory setting should be less efficacious than the newly diagnosed setting. As with most new treatments, imatinib was first tested in the clinical setting of patients who were resistant or refractory to interferon-based therapies, the gold standard treatment at the time (when stem cell transplantation was not possible). This group of studies provides information on imatinib efficacy in the treatment resistant and refractory setting, timing of best imatinib response, duration of response (PFS), survival (OS) and dose response.

Several phase II studies of imatinib for interferon resistant or refractory disease exist (Table 3). In the first major published imatinib clinical trial, Druker and colleagues demonstrated that imatinib had activity in the interferon resistant or refractory setting, documenting Major CRs of up to 50 percent.66 This was a landmark study, establishing that an oral targeted therapy could have dramatic activity in a disease resistant setting.

Kantarjian, Sawyers and colleagues conducted the largest phase II open-label study.2 Efficacy estimates from this study of 400 mg daily with a median duration of imatinib treatment of 17.9 months indicated that the Major CR rate for the interferon resistant or refractory group of patients was 60 percent with a Complete CR rate of 41 percent. The imatinib dose was increased to 800 mg when patients had not achieved a CHR by 3 months, a Major CR by 12 months, or relapsed after CHR. These estimates have been pretty consistent across this entire group of studies.8 Patients treated earlier in their course (early CP, i.e., <1 year since diagnosis) have fared better than those whose disease is in late CP (>1 year since diagnosis), with a 62 percent Complete CR rate for early CP and 41 percent for late.44 The estimates were slightly higher than reported for interferon resistant or refractory CP in the 2002 FDA approval summary (Major CR 31 percent, Complete CR rate 13 percent).82

Reasons for needing to change from interferon-based therapy varied, and included resistance to the medication (failure to achieve the desired response within a defined timeframe), relapse (return of disease after response has been achieved), and intolerance (non-hematologic > Grade 3 toxicity). Patients with hematologic or cytogenetic relapse after interferon-based therapy had higher response rates to imatinib than those with resistant disease after 6 months of therapy (cytogenetic relapse after interferon, 76 percent Complete CR with imatinib; cytogenetic resistance to interferon, 31 percent Complete CR with imatinib). Patients who were interferon intolerant had intermediate response rates, however this group was older than the other patient participants (50 percent with age >60 vs. 40 percent for rest of participants), consistent with the fact that they were not tolerating the side effects of interferon well.

As observed with the newly diagnosed group, cytogenetic responses continued to accrue after up to 12 months of imatinib therapy.8,73 Periods after 12 months were not reported. Importantly, though, patients who achieved any CR early (i.e., by 3 months) had substantially longer PFS and OS (OS if achieved Major or Minor CR by 3 months, 95 percent, if not 72 percent, p<0.0001).72 Similarly, an early Major CR that is achieved by 6 months was also associated with longer PFS and OS (OS if achieved Major CR by 6 months, 95 percent, if not 78 percent, p=0.001). Extending these findings to molecular responses, Rosti, et al. report that overall survival is better is a major molecular response is achieved.8

Because the interferon resistant or refractory is the oldest group of longitudinal studies of imatinib in CML, the longest survival followup data are available for this group of patients. Duration of response and survival are reflected in the 4-year followup study by Kantarjian, Cortes, and colleagues.72 Among their full cohort of 261 patients they described a 4-year OS rate of 86 percent and PFS of 80 percent. They compared these PFS rates to a matched cohort of historical controls under treatment at their institution from 1982 to 1997. The historical cohort had a 4-year PFS of 43 percent (compared to the imatinib cohort p<0.0001).

Appropriate dosing continues to be a question. Phase I dose ranging studies with phase II outcomes correlates demonstrated that doses in the 500 mg range were most efficacious.66 Some patients resistant to lower doses of imatinib achieved a response when the dose was increased to 800 mg .71,74 As expected, response rates were lower (Complete CR 5-19 percent) and less durable (43 percent with loss of response by 416 days). Similarly, increasing the dose could overcome relapses, such that patients who relapsed at 400 mg imatinib could still achieve a cytogenetic response when the dose was increased to 800 mg (18 percent Complete CR).71 Finally, Cortes, et al. reported 89 percent Complete CR rates when the initial imatinib dose was 800 mg, although only 33 participants were involved in this study.69

One challenge for this group of studies is that there are several publications presenting data from different combination of the same group of patients.2,70-72 These patients were recruited through several phase II industry-sponsored trials (Novartis 110, 112, 113) and the various publications represent different clinical questions, analyses, comparison groups, and followup periods. There is a risk of misinterpreting these as multiple independent datasets corroborating the efficacy estimate.

Chronic phase — Previous stem cell transplant and heavily pretreated

As the number and intensity of previous treatment increases, there is progressive decrease in the chance of response to new treatments. A critical question for imatinib is whether it is an option for patients who have become resistant to multiple prior therapies, and whether it precludes other subsequent therapy. Of particular interest is stem cell transplantation (SCT, a.k.a., bone marrow transplantation), which includes intensive myelosuppressive cytotoxic chemotherapy often with multiple agents. Allogeneic transplant also carries a substantial risk of graft versus host disease (GVHD).

Cervantes, et al. demonstrated that 400 mg of imatinib yielded substantial Complete CR rates for 33 patients with prior autologous SCT (33 percent at 12 months), similar to that of the comparison sample of 65 interferon refractory patients who had not had a transplant (38 percent; Table 4).37 Similar response rates were substantiated across this group of trials, with some studies noting even substantially higher Complete CR rates (33-85 percent).

Accelerated phase

CML that is more advanced at presentation has a poorer prognosis. Imatinib is still efficacious in the accelerated phase setting, as demonstrated by Talpaz, et al. (Table 5).87 Complete CR rates ranged from 11 to 19 percent, with the 600 mg dose being more efficacious than 400 mg. In a subsequent followup abstract, the Major CR rate was 48 percent with a median followup of 38 months and the median survival had not been reached. A group of historical controls (accelerated phase, not otherwise described) were reported to have a median survival of 21 months.88 The 3-year OS was estimated at 53 percent.

Blastic phase/blast crisis

During the chronic phase there is massive clonal expansion of CML cells. In the blastic phase the cells lose the ability to differentiate and the leukemia advances rapidly. Blastic phase CML has the poorest prognosis with an expected survival of 3-6 months. Historically it has been poorly responsive to any therapy. Median survival is 21-29 weeks, even with aggressive acute leukemia treatment plans.141 Database review studies have indicated a 10-year survival after bone marrow transplantation of 0 percent (1996 report).142

Imatinib has been shown to have efficacy in the blastic setting (Table 6). Sawyers, et al. report the largest Phase II trial involving 260 participants with a median duration of treatment of 4 months.3 A total of 31 percent had a sustained CHR for over 4 weeks and 7 percent had a Complete CR. For those who did respond to imatinib, the estimated median response duration was 10 months. OS was estimated as 6.9 months (95 percent CI, 5.7-8.7 months) with 43 percent survival at 9 months and 20 percent at 18 months. In all studies that evaluated response by blast type, lymphoid blast crisis had better response rates than non-lymphoid (myeloid) blast crisis.89,90 Previously untreated patients always had a better response than those who were previously treated.3 Doses ranged from 400-1000 mg without a clear pattern for maximal efficacy. Sawyers, et al. started with 400 mg and increased to a maximum of 800 mg when the disease was resistant or refractory.3

Additional efficacy tables

Table 7 reviews three reports that presented response to imatinib across phases. Two reports were summed from the three large phase II Novartis trials submitted to the FDA as part of the 2002 imatinib approval process.82,83 These studies are instructive in that they provide validation of the differential effect of imatinib therapy by phase of disease and the efficacy estimates previously presented, as well as additional estimates of treatment durability. With a median followup of 40 months 64 percent of CP participants were still taking imatinib.83 Among CP patients with Major CR, 82 percent were still on imatinib at 3 years, with PFS 80 percent and OS 88 percent. For AP and BP the 3-year PFS was 55 percent and 5 percent, respectively. A third study conducted with 128 patients who had a prior allogeneic SCT also validates the differential effect of imatinib by phase and the activity of imatinib in the heavily pretreated post-allogeneic SCT setting.84 The overall and CP Complete CR rates of 42 percent and 58 percent described were consistent with the previous group of allogeneic studies.

Table 8 presents two additional studies that did not naturally fit into the other tables. Both of these were preliminary trials assessing the tolerability and efficacy of drug combinations including imatinib in newly diagnosed CP CML. Gardembas and colleagues described imatinib combined with cytarabine38 and Baccarani, et al. reported imatinib plus pegylated interferon.91 Both trials reported Complete CR rates that were no better than those seen in the IRIS study with imatinib alone. Interpretation of these trials is limited by the shorter followup periods; additional cytogenetic and molecular responses may accumulate with time making these combination therapies more interesting as the data mature.

Quality of Life

Quality of life (QOL) is another important efficacy outcome. Hahn and colleagues investigated the QOL of newly diagnosed CP patients receiving imatinib vs. interferon plus cytarabine in the IRIS study.60,61 The Functional Assessment of Cancer Therapies-Biologic Response Modifiers (FACT-BRM) instrument was used.143 The primary QOL outcome was the Trial Outcome Index (TOI; 27 items, score range 0-108) and secondary endpoints included social/family well-being (SFWB; 7 items range 0-28) and emotional well being (EWB; 6 items, range from 0-24). Higher scores indicated better QOL. Quality of life was measured at baseline, monthly for 6 months, then at 9, 12, and 18 months. Imatinib treated patients scored significantly higher on all of these QOL measurements. The mean TOI across the trial was 84.4 for imatinib treated patients and 67.7 for patients on interferon plus cytarabine (p<0.001). Patients on the interferon plus cytarabine arm had a substantially greater decrease in TOI across time than those on imatinib. This work was recently repeated in a phase II study conducted by Pasquini and colleagues in Brazil.76 Imatinib led to clinically significant increases in TOI at 1, 6 ,and 12 months.

Adverse events

Table 10 reviews the adverse events reported across the studies. In the IRIS trial, imatinib most commonly caused neutropenia (61 percent), thrombocytopenia (57 percent), superficial edema (56 percent), nausea (44 percent), and abnormal liver function results (43 percent).59 Interferon plus cytarabine most commonly caused thrombocytopenia (79 percent), abnormal liver function results (74 percent), neutropenia (67 percent), fatigue (66 percent), nausea (61 percent), anemia (55 percent), and headache (43 percent). The incidence of grade 3/4 side effects was primarily hematological with imatinib (neutropenia 14 percent and thrombocytopenia 8 percent) whereas interferon plus cytarabine included fatigue (24 percent) and hematological (neutropenia 25 percent and thrombocytopenia 17 percent). The incidence of side effects increased with imatinib dose and phase of illness, with hematologic side effects particularly increasing with advancing phases of illness.

In addition to the adverse events commonly described across this group of studies, four individual reports of adverse events were identified. Valeyrie and colleagues prospectively followed 54 patients started on imatinib.94 Eighty-nine percent experienced at least one cutaneous reaction; 67 percent had rashes, 65 percent edema and 41 percent pruritis. Six percent had severe enough rash to discontinue therapy either temporarily or permanently. The rate of rash increased with imatinib dose. In a similar study of 78 patients by Drummond, et al., 12 percent of patients had rashes that could be directly attributed to imatinib.92 Steegmann reported a prospective study of gamma globulin levels in 36 patients receiving imatinib for CML when resistant to or intolerant of interferon. Low serum IgG, IgA, and IgM levels were identified in 28 percent, 14 percent and 22 percent of patients, respectively.93 Finally, Al-Ali and colleagues identified that imatinib caused elevated creatinine kinase (CK) levels of >50 percent above baseline in 81% of the 113 patient cohort studied; elevation was highest for those who reported cramps or myalgias.95 Patients whose CK levels were elevated after 6 months of imatinib had higher rates of Major CR (p=0.048).


Ideally, treatment is matched to those patients most likely to respond to that treatment. Certain clinical and molecular characteristics can be used to predict which patients with CML are more or less likely to respond to imatinib. These predictors of response to imatinib are distinct from the disease characteristics that correlate with prognosis irrespective of treatment plan. For example, the most important prognostic factor is the phase of disease. Some prognostic factors are also associated with response to treatment. Clinical characteristics predicting response were presented in the Efficacy section and included:

  • Phase of disease (CP, AP and BP; early vs. late CP).
  • Previous treatment before imatinib (interferon, stem cell transplantation).
  • Reason that previous treatments were discontinued (resistant, refractory, intolerant).
  • Dose.

Many authors have reviewed the correlation between clinical prognostic factors (e.g., splenomegaly, percentage of blasts in the peripheral blood, platelet count) and tumor response or survival with imatinib. As expected, most of the known prognostic factors can be used to identify high risk and low risk patients in the setting of imatinib therapy in a similar manner to other treatment settings. A full review of the hazard ratios for these clinical prognostic factors is outside the scope of this review. Here we concentrate on molecular factors that predict response to imatinib and are likely to be related to the targeted action of the drug.

The molecular predictors can be arbitrarily divided into four groups. The first three groups are based upon whether the assessment focuses on genetic material (DNA), production of the RNA message, or the tyrosine kinase protein and its interaction with imatinib. A fourth group includes other miscellaneous predictors. Group 1 includes DNA predictors are related to the formation of Ph, the evidence of impact of imatinib on the Ph, the accumulation of other DNA abnormalities within the CML cells, or genetic profiling to predict imatinib responders. Group 2 includes RNA predictors that relate to the production of the BCR-ABL mRNA transcripts including trends in production over time. Group 3 relates to changes in the tyrosine kinase protein that influence the activity of imatinib. Group 4 includes other related predictors that were identified in this review such as bone marrow cellularity and myelosuppresssion. These groups can be further divided into characteristics identified at the start of imatinib therapy and characteristics that can be evaluated during therapy to predict response (subclassification A or B).

Assessment of study quality is reviewed in Chapter 2. Quality scores reflect study reporting quality from a clinical research standpoint, not the quality of the basic science. In a broad review of the literature such as this one, it is difficult to determine which predictors have been exhaustively scientifically validated and which ones are only investigational. The volume of studies citing an individual predictor is used as a proxy indicator. These tables have been arranged so that potential predictors with a large number of supporting studies are cited at the beginning of the tables and emerging predictors are cited at the end.

Molecular predictors: Group 1A—DNA factors at the start of imatinib therapy

Ph+ cells measured during cytogenetic analysis is a measurement of burden of disease. This is represented in terms of "percentage of Ph+ metaphases" at the start of imatinib therapy. Five studies evaluated the relationship between this predictor and tumor response, progression, or survival. There were significantly more patients with a Major CR when <90 percent of metaphases where Ph+ at the start of therapy.1,2 A similar trend for survival was seen, but not statistically significant.1 There were significantly more patients with a Complete CR when <100 percent of metaphases where Ph+ at the start of therapy; overall survival was longer too.75 For those patients increased to 800 mg of imatinib due to disease resistance at 400 mg, complete and partial cytogenetic response were again more likely if Ph+ cells represented <100 percent of metaphases.74 In terms of disease progression, there was not a statistically significant relationship between CML hematologic relapse and those patients with >98 percent Ph+ metaphases at the beginning of therapy, but the trend for relapse followed that previously seen.96 These secondary analyses were predominantly from studies of patients with CML in chronic phase that is resistant or refractory to interferon (CP-IFN-r); one study included other CP patients and AP patients.1 In general, patients with a smaller burden of disease at the start of imatinib therapy were more likely to have a Major CR, Complete CR, and/or improved overall survival.

Chromosomal abnormalities in addition to the Ph have been repeatedly investigated as a potential prognostic and therapeutic predictor in CML. Cytogenetic abnormalities have been investigated both at the time of initial diagnosis and with clinical disease progression (e.g., from chronic to accelerated phase). The language that various authors use to describe this process is imprecise, including descriptions of "other chromosomal abnormalities," "complex cytogenetics," and "cytogenetic clonal evolution". Overall, the most common terminology in "clonal evolution" and therefore this grouping will be used to represent this category of predictive markers.

Clonal evolution at the time of initial diagnosis may be a marker for more advanced or aggressive disease. Indeed, larger studies of patients in AP and BP supported that clonal evolution at baseline predicted poorer survival (p<0.005)3,4 and likely predicted disease progression (p=0.086).87 Smaller studies did not support these findings.67,90 Cytogenetic clonal evolution is often a hallmark of CML as it progresses from chronic to more advanced phases. Similar to phase being a clinical predictor of response to imatinib, clonal evolution may be a molecular predictor. Ten studies including patients in CP and CP-IFN-r considered cytogenetic clonal evolution as a predictor of tumor response, although it was likely that these studies reflected multiple presentations of the same patient populations. Taken together these studies suggested that cytogenetic clonal evolution inconsistently predicted disease response,1,2,70,75 but was a major predictor of the risk of disease relapse (relative risks (RR) reported 4.34, 4.912, and 14.8)96,97,99 and survival.1,44,70,72,75

CD34 is an antigen that is selectively expressed on myeloid and lymphoid hematopoetic progenitor cells. Marin and Elliot both presented abstracts that indicated that the percent of CD34+ cells in the bone marrow in CML correlated with tumor response.100,101

Variant Ph translocations occur in up to 10 percent of cases of CML. The variant Ph may lead to variant BCR-ABL tyrosine kinase proteins and therefore affect imatinib's efficacy. Prior to the era of imatinib, variant Ph was not associated with prognosis except perhaps abnormalities involving chromosome 17.49 In an analysis that included patients in CP and AP, El-Zamaity and colleagues did not identify a significantly shorter duration of response with variant Ph as compared to other patients with CML.97

Deletions of the resultant DNA on chromosome 9 can be seen in up to 15 percent of cases of CML.50 Chromosome 9 deletions are known to negatively affect prognosis, decreasing survival by up to 20 percent at 5 years.53,54 These studies were conducted predominantly in patients on interferon-based therapies.50 In the setting of imatinib, chromosome 9 deletions lead to poorer PFS in CP, AP and BP settings (p=0.02).51 Another study found no differences in major CR or complete CR in CP patients.144 Overall survival was not significantly different in either study with median followup of 48 months. Longer periods of followup may be needed.

Investigations of genetic patterns are underway. A number of genes are known to be related to drug resistance and programmed cell death (apoptosis) in leukemic cells. Evaluation of gene expression suggested that MRP-1 was overexpressed in blast crisis CML, and that MRP-1 overexpression was significantly correlated with poor tumor response to imatinib.102 Using gene microarray techniques, McLean and colleagues identified a genomic profile and microarray pattern characteristic of tumor response in CP CML. Patients whose CML met this ideal microarray profile had a substantially greater likelihood of Complete CR (odd ratio (OR) 200, 95% CI 19-3096) and Major CR (OR 19.9, 95% CI 6-67).103

In summary, DNA factors at the start of imatinib therapy that predict poorer tumor response and/or survival include the following:

  • 90-100 percent of metaphases are Ph+ at the start of imatinib.
  • Clonal evolution in AP or BP.
  • Clonal evolution in CP (predicts risk of relapse and poorer survival).
  • Higher percentage of CD34+ cells in the bone marrow.
  • Chromosome 9 deletions.
  • Genetic profiles.

Molecular predictors: Group 1B—DNA factors monitored during imatinib therapy

Cytogenetic response (CR) is the most commonly used surrogate marker of tumor response for CML. Its relationship to PFS and OS in the setting of imatinib therapy has been confirmed by at least seven studies involving all phases of CML.8,62,68,72,75,83,96 Timing of the CR is also important. Across the analyses that evaluated the time course of the CR, CR by 3 or 6 months strongly predicted PFS and OS.62,72,75 In the only study that compared timepoints, partial CR by 6 months was most predictive of survival.72

Similarly, the degree of reduction in CD34+ cells in the bone marrow can be considered another surrogate marker of tumor response. Marin demonstrated that the degree of reduction of CD34+ cells in CML in the setting of imatinib treatment correlated with progression free survival (RR 0.88, 95 percent CI 0.53-0.93).100 This is consistent with imatinib decreasing the percentage of blasts and normalization to a CHR.

In summary, DNA factors monitored during therapy that predict better tumor response and/or survival include the following:

  • Cytogenetic response and
  • Degree of reduction of CD34+ cells in the bone marrow.

Molecular predictors: Group 2—Production of the RNA message

All of the RNA factors identified related to quantification of BCR-ABL mRNA transcripts and evaluation of their time course using Q-RT-PCR. All phases of CML were studied. A decrease in mRNA transcripts with treatment is a "molecular response" (MR), and is a surrogate marker of CML tumor response. RNA factors can also be considered in terms of evaluation prior to initiation of imatinib therapy and then followup evaluation during treatment.

Nine studies support the association between MR and overall tumor response.5-12 An individual patient's best MR predicts survival and those with very low levels of residual disease (median ratio <0.1 percent) have the more durable Complete CRs.10 Among all patients in the IRIS study who achieved a Complete CR, those who received imatinib had a greater MR than those who received interferon plus cytarabine (p=0.036).11

Response to imatinib is independent of BCR-ABL mRNA transcript number at the start of treatment.6,55,104 However, molecular monitoring during imatinib therapy is predictive of overall tumor response. Generally, this is considered in terms of transcript level or log reduction in transcript levels at 1, 3, 6, or 12 months. Median log reduction of > 2 at both 3 and 6 months was predictive of continued tumor response at 24 months.33 Median log reduction of > 3 at 12 months was also predictive of continued tumor response at 24 months.11 Similarly, when the BCR-ABL/ABL ratio is <50 percent at 4 weeks, the PFS at 500 days is 100 percent, vs. 45 percent for those who do not achieve a ratio of <50 percent.48 Based upon data from the IRIS study, BCR-ABL transcript levels did not decrease substantially after 24 months on imatinib treatment.8,113 There are also substantially more IRIS patients that received imatinib with >3 log reductions in transcript levels than those who received interferon plus cytarabine.11

In summary, factors related to production of the RNA message that are monitored during therapy and predict better tumor response include the following:

  • Molecular response.
  • > 2 log reduction in BCR-ABL mRNA transcripts at 3 or 6 months.
  • > 3 log reduction in BCR-ABL mRNA transcripts at 12 months.
  • BCR-ABL/ABL ratio <50 percent at 4 weeks.

Molecular predictors: Group 3—Interaction between the tyrosine kinase protein and imatinib

Mutations in the tyrosine kinase protein have been an active area of inquiry. Only three studies met the eligibility requirements for this review and were therefore included on Table 14.55,108,145 These studies were of lower quality than the majority of included articles, mainly because they were basic science reports with minor clinical correlations. Since they focused on the basic science, there was less attention in the manuscript to the traditional quality reporting items that are usually considered during secondary clinical research summaries. Further, several studies did not meet the explicit criteria for this review and therefore were highlighted within the "future directions" studies only (Table 1d, "Mechanism of action"). These studies were excluded primarily because they did not clearly provide quantitative assessment of the correlation between the molecular findings and response to imatinib. Taken together, the group of studies presented in Tables 1d and 14 suggest that there is substantial current research effort focusing on the molecular mechanisms of imatinib resistance at the protein level. Some of this work focuses on the gene expression corresponding to imatinib resistance, such as the MRP-1 studies described previously. Others evaluate the relationship between mutations in the tyrosine kinase domain that lead to changes in the protein which might confer imatinib resistance.[Shah, 2002 #292; Hochhaus, 2002 #285; Soverini, 2004 #858] Of particular interest are mutations in the p-loop of the protein where ATP binds and the protein pocket where imatinib binds.112,114-116 These data are in development; clear evidence of the clinical utility of such information for predicting tumor response and overall survival with imatinib is not available yet.

White and colleagues described an in vitro assay to predict imatinib's ability to inhibit phosphorylation of the adaptor protein Crkl.145 Crk1 binds BCR-ABL directly and plays a functional role in BCR-ABL-mediated transformation to cancerous CML cells by linking the kinase signal to downstream effector pathways.146 Previous in vitro studies have shown that Crkl phosphorylation correlated with untreated disease and relapse after imatinib, while lack of phosphorylation correlated with response to imatinib.146 White, et al. measured in vitro levels of Crkl phosphorylation of the patients CML cells in the setting of imatinib; using a scoring system of high and low levels of Crkl phosphorylation measured by the IC50, they correlated the IC50 to Major MR. Among newly diagnosed CP CML patients, low IC50 at diagnosis correlated with ability to achieve a Major MR at 12 months. This correlation was particularly strong for those patients with low Sokal scores.

In summary, protein factors related to the interaction between the tyrosine kinase protein and imatinib that can be monitored during therapy and that predict better tumor response include the following:

  • In vitro evidence of imatinib's ability to reduce Crkl phosphorylation.

Molecular predictors: Group 4—Other factors

Several other molecular studies are presented in Table 15. Bone marrow cellularity decreases when CML responds to imatinib, an expected finding.109 Myelosuppression due to imatinib of > Grade 3 predicts poorer Major MR rates with imatinib, and if the myelosuppression persists for > 2 weeks the chance of Major MR is even lower.110

The concept of "cure" and complete disease eradication in CML is murky. Even when patients are in CCR, evidence of CML can be found. Bhatia and colleagues showed that all of the 15 patients in Complete CR studied had evidence of BCR-ABL in their CD34+ cells as identified by FISH or RT-PCR up to 61 months after starting imatinib.27 O'Dwyer reported similar findings for seven patients in Major CR.35 Using sensitive RT-PCR techniques Paschka, et al. found evidence of BCR-ABL in all samples of CCR patients on imatinib.10 Taken together, these data support the notion that complete remission in CML may be conversion to a low grade chronic disease with continuous potential for relapse over the long term. Using the previous definition from the transplantation literature that "cure" is continued Complete CR at 5 years,13,34 "cure" may be a relative state of disease control rather than complete eradication.

In summary, other factors monitored during therapy that predict poorer tumor response include the following:

  • Myelosuppression due to imatinib of greater than Grade 2.
  • Myelosuppression persisting for more than two weeks.

Return to Contents
Proceed to Next Section

The information on this page is archived and provided for reference purposes only.


AHRQ Advancing Excellence in Health Care