Saturday, May 07, 2011

The SEER-Medicare data on CLL

One of the problems about clinical trials is that they seldom address the disease where it is at. Most patients with CLL are over 70 and have co-morbidities. Such patients are seldom entered into clinical trials. Ideally we should be doing trials in this group of patients, but it ain't gonna happen.

In the Blood of March 31 an article by Danese et al tried to help with this problem. With his co-workers he examined the SEER-Medicare database for how patients with CLL were treated. Although I have waxed lyrical about the German CLL8 trial, it has to be admitted that the median age of entry to this trial was 61 and most patients were of high performance status. Although generally CLL8 demonstrated the benefit to adding rituximab to FC, for patients who were older than 70 years, the numbers were too small to show any benefit. It was also not possible to show benefit for patients with stage C disease.

The SEER-Medicare study were able to look at a large group of CLL patients from defined areas and characterize their initial use of infused therapy (chemotherapy and rituximab) and to evaluate the outcomes in patients using different types of infused therapy. One of the drawbacks of the study is that it is unable to make any comment about the use of oral therapy like chlorambucil so that those apparently receiving rituximab alone might have been receiving chlorambucil plus rituximab or even oral fludarabine and rituximab.

Patients registered between January 1999 and December 2005 were studied. 6433 patients were identified with an average of 40 months follow-up. The median age at diagnosis was 77 years and 1675 were of advanced stage (had anemia or thrombocytopenia. No allowance was made for immune or other types of cytopenia though 3.8% were known to have hemolytic anemia). The SEER population differs slightly from the overall population of the USA in being more urban with more ethnic minorities.

2040 patients received infusional therapy, 1429 receiving chemotherapy alone and 319 rituximab alone. 292 received both, of whom 95 received flufarabine and most of the rest cyclophosphamide. Only 53 patients were identified as receiving FC plus or minus rituximab. Lower rates of infusional therapy was seen in patients older than 80 years, those with more co-morbidities, women, racial ethnic minorities and those residing in poor neighborhoods. Median survival was 52 months for those receiving rituximab with chemotherapy compared with 34 months for those receiving chemotherapy alone. Those receiving rituximab alone had a median survival of 53 months.

There are all sorts of ways of interpreting these figures. We have no data on prognostic factors and no way of matching those having chemotherapy alone and chemotherapy plus immunotherapy patients. We might reasonably expect that those who got rituximab alone were not such severe cases but we do not know. One explanation might be that adding rituximab does indeed enhance survival just as CLL8 says it does and this translates into the ordinary rather than trials population. In this study the advantage continued for the elderly but was not present for those with significant cytopenias. It was greatest when compared with those who had fludarabine rather than other types of chemotherapy.

One other thing of note was that the use of rituximab increased over this period despite the fact that at no time was it licensed for use in CLL. remember these patients were all Medicare patients.

3 comments:

Grateful said...

In addition to the lack of prognostic data because Medicare does not pay specifically for prognostic tests, another problem is that Medicare uses ICD9 codes, which do not include a specific code for a diagnosis of CLL. Instead, ICD9 has a code for chronic LYMPHOID leukemia, which includes but is not limited to CLL.

Terry Hamblin said...

Yes that is another flaw in the data.

Anonymous said...

Both of you are wrong because it is obvious that you don't know the design of SEER-Medicare linked data. All CLL cases were identified from the data of SEER cancer registries not by ICD-9-CM diagnosis codes.