Prediction of response to imatinib by cDNA microarray analysis,
R Ohno, Y Nakamura,
Seminars in Hematology, April 2003, 40.

Discussed by Eva

Finally, something in print on the aspect of predicting by gene expression profiling using the cDNA microarray a leukemic patient respone to imatinib.

The first abstracts on this had appeared in ASH 2002 and was a Novartis and IRIS study  ([527] Correlation of Cytogenetic Response with Gene Expression Profiles in Chronic Myelogenous Leukemia Patients Treated with Imatinib (Gleevec™/Glivec®).
Lee Anne McLean, Insa Gathman, Renaud Capdeville, International Randomized Study of Interferon vs. Glivec (IRIS) Study Group, Mihael H. Polymeropoulos, Marlene Dressman (Intr. by Laurie A. Letvak). Pharmacogenetics, Novartis Pharmaceuticals Corp., Gaithersburg, MD, USA; Oncology, Novartis Pharma AG, Basel, Switzerland), and the results there were sort of similar to the general findings of this Japanese study.  This paper also reviewed the only publication to date on use of gene expression in predicting response to matinib for Ph+ ALL, apaper  which I had n the past wanted to grab, so I was quite pleased to see it discussed here, as well.

Why is there a sort of pressing  need to develop another  prognostic tool such as cDNA array or gene expression profiling (GEP) in CML ?

The phenomenal cytogentic responses to IM are well known, however, the drug is still new to be able to really be validated as curative (and now we do know of resistance issues in some situations with this drug). The only known curative treatment as we have discussed is an allogenic BMT, which has a considerable morbidity risk.  People have been responding to IM and as  discussed in patient group lists and booards  there remains that niggling question of having made the right decision to stay on with IM and foregoing the BMT route early on, if in the end we experience resistance and non reponse to IM later on, when our chances for BMT become more riskier with time.

Such dilemmas and questions bothering CML patients, may be best addressed if one had an experimental tool that could be used to predict whether we would be responding to IM well, or whether we had the potential to relapse or resist IM in the future, so that as early as possible patients could already be guided towards  seeking other therapies like a BMT or begin combination drugs or other drugs. This tool if successful in predicting response to Imatinib, could also further be used in developing a personalized approach to cancer therapy, whence one's personal cells can be exposed to new drugs in vitro, and from their gene expression results indicate the possibility of success and minimization of adverse side effects.  To do the latter for personalized treatment, one should however also include the SNP or single nuccler polymorphisms considerions which can impact on th metabolism of the drug.  The prevalence of this technique and translation to clinical settings will make choice of clinical trial participation more rational and less of a hit and try approach.

So where they able to find genes which could discriminate CML patients who would respond to IM and those wo would not ?

Unlike the Novartis Maryland study  ( using 12 month period on IM) which appeared in the ASH Abstract in Dec, the Japanese study made use of only a 12 week or 3 month period on IM. Their definitions of responsiveness and non-responsiveness were thus a bit different as yet no to few CCRs could yet be attained within that time interval nor relapses be observable in such a short time frame. 

Back to CML Heads Up !

Nevertheless this time period used in the study is quite important, as one would like to make that decision to go for a BMT or some other course  sooner than possible, and to some a decision past three months seems ideal (though I know this is debatable as some institutions go for a decision at 6 months or prior the 1 year post dx date).  We will have to await the full report of the Nvrts-IRIS study as that one had more cases used and had the potential of a retrospective study with longer IM treatment follow-up for potential CCR correlations . The number of genes used in the array were in the same order 10 to the 4 genes, and about the same number of genes were found to be discriminatory of  responderr and nonresponders.  In this Japanese study, 30 discriminating genes were found from the 79 predicted ones. Whereas 31 genes were identified in the IRIS study.

Both groups used their discriminatory genes to check or validate if it was indeed ableto predict response . In this Japanese study they tested in only 4 other patients that were not  included in their candidate gene search studies. In the IRIS study,it was not quite clear to me if their test was done on patients belonging to the same study set. Both groups claim, success in the use of cDNA discriminatory genes in predicting response but did caution that a larger patient population be tried for validation. Wile the IRIS study recommended use on a different or independent population set for validation.

So what about the use of  similar approaches  in Ph+ ALL response to IM prediction ?

Hoffman et al.group study was reviewed. This study used only 10 to the 3 genes. Their study also showed success in determining 95 candidate genes for discrimination, with 56 genes showing high levels of discrimination and prediction of response to Imatinib. Curiously, the genes in this study did not overlap with the genes of the Japanese study. (And it will be a curiosity if the IRIS study list of genes will turn out to be the same as that of the Japanese study. Corroboration studies is urgently needed for this tool to gain wide acceptance in the prediction of drug response.)

Have there been other studies employing cDNA arrays or GEPs in other malignancies ?

So far only the  Ph+ ALL by the Hoffman grp has been published. Here we have the   Im response in CML by this Japanes group though we are awqre of the IRIS study presented at ASH. And scouring through the conferences and the literature one sees a lot of GEP studies being employed for other drugs in various cancers.

Why has this been done mainlyon leukemias and not for solid cancers ?

One needs to have  pure colony of abnormal cells with no contamination from normal cells, a task  easier done for blood cancers than  for solid tumors.

The Japanese group thus setout to do a similar response prediction study in AML for the known induction therapy of idarubicin and cytrabine using the same number of genesi n the array.  They also found 28 genes which were used to discriminate good and nonresponders from samples prior treatment, and these prediction to sensitivity correlated well with achievement of actual responses and eventual remissions. However 4 out of their 44 patients predicted  to respond did not achieve complete remissionwhle 3 who scored to beonresponders achieved remission.

The imperfection in the prediction in AML was attributed to host (staus of patient) and disease stage (wbc and LDH levels at diagnosis) factors which they claim impact response but however are not  reflected in gene exression profiles of leukemia cells.  Furthermore, two drugs are in use for induction in acute leukemias would beharder to predict than thesingle drug use of imatinib in CML. (I believe they have high hopes of perfect prediction when larger populations of CML patients will be used to validate their 30 discriminatory genes.)  In contrast the IRIS study said they had  ". Overall, the set of predictor genes was able to correctly predict cytogenetic response status in 94% of the patients (62 of 66). Three patients were incorrectly predicted to be NoCR and one patient was incorrectly predicted to have CCR. "

There was a discussion on the dicriminatry genes that were identified, and Imay post on those some other time.


Copy of article by subscription

Search the mariners faq and glossary
database

Join us in an on-line community

Join d discussion forum

Cml ahoy !