The SEER database is a wonderful tool for understanding cancer and the recent release of hematological data from 1973 to 2004 has prompted several studies including one in last month's Blood by Brenner et al from Heidelberg, Germany and Cornell, New York. The suggestion in this paper is that long term survival expectations of patients with CLL have substantially improved for patients over the past two decades.
Without wishing to discourage readers I am suspicious that this claim is at best exaggerated.
The SEER database is assembled from cancer registries in Connecticut, New Mexico, Utah, Iowa, Hawaii and Atlanta, Detroit, Seattle-Puget Sound and San Francisco-Oakland. It takes in about 30 million people. There were 20,491 cases of CLL diagnosed between 1974 and 2004, 12,120 men and 8371 women. There was a steady increase in the number of cases: 3642 between 80/84, 3969 between 85/89, 4236 for 90/94, 4191 for 95/99 and 4454 for 00/04. The changes in the age groups was rather strange. In the period 80/84 to 00/04 there was a 44% increase in the under 60s, a 3% reduction in those aged 60-69, an 18% rise in those aged 70-79 and a 41% increase in the over 80s.
The percentage surviving 5 years has increased over the same period by 6% and the percentage surviving 10 years has increased by 7% The improvement has been seen in both sexes and at all ages except that the 10-years survival if the over-80s has not improved at all.
They have also looked at relative survival, which means how did they do compared to people of the same age and sex who didn't have CLL? Again, here things seem to have improved from 69% to 75% at 5 years and from 46% to 55% at 10 years, but again for the over 80s there's been no improvement.
Now one explanation is that treatment has got a bit better (though with the over 80s only 30% as likely to survive for 10 years as the non-CLL population, things are not good).
However, there have been other changes over the past 20 years. Supportive care has improved, for example. It may be that we are better able to keep people alive who have CLL even if we can't do much to treat the disease.
Cancer Registries are notoriously poor at collecting cases of CLL. Patients are often not admitted to hospital, and there is usually no histology. Furthermore, the fact that the patient lived with CLL that never caused him any problems is often omitted from death certificates.
But I think that the biggest difference has been in the diagnosis of CLL. Immunophenotyping has meant that patients with lymphoma with blood spillover are no lnger diagnosed as CLL, and generally these patients had a worse prognosis than cases of CLL. But overshadowing all these is earlier diagnosis. It is interesting that patients under the age of 60 are more frequently diagnosed now. This might be because the disease is appearing at a younger age, but it also might mean that the disease is picked up earlier in its presymptomatic course. Patients might only survive longer now because they are diagnosed earlier. I believe this is very likely to be true. When the study started you needed a lymphocyte count of 15,000 to diagnose CLL, but the rules have changed and now you only need 5,000.
I'm sorry but I find it hard to believe there hasn't been some increase in life expectancy for CLL patients.
ReplyDeleteI am a case in point. I was quite sick before treatment, transfusion-dependent, skyrocketing lymphocyte counts (doubling every two-three months or so). Treatment knocked everything back and I had a complete remission that has lasted about two years so far.
I don't believe you could say that not treating my CLL would have had such an effect. Chlorambucil was ineffective.
FCR and transplants do lead to long-lasting remissions that otherwise would have ended in death. I don't see how anyone could conclude otherwise, especially in the case of transplants.
I am not saying that nobody has benefited from modern treatment; of course they have. What is clear from the SEER statistics is that the number surviving 5 years has only gone from 54% to 60% ad the number surviving 10 years has only gone from 27% to 34% in the past 20 years, and that there are other explanations for this finding. It would be interesting to look at British statistics where chlorambucil was the main drug for most of that time, rather than fludarabine.
ReplyDeleteBecause of the immunosuppressive side effects of anti-CLL drugs it is possible that more patients are getting good remissions, but that this is not translating into more 10-year survivals because of the rate of complications.
Transplants are a case in point. They certainly lead to long term remissions in some patients and some are cured, but until recently transplant related mortality was 40%.
As a gastroenterologist I have noted as teady rise in the incidence of Crohns Disease throughout my career. In addition the incidence of several tumors of the gastrointestinal tract has clearly changed, with a definite increase in adenocarcinoma of the esophagus (likely due to our use of antibiotics, which kill of H. pylori in the stomach) and pancreatic cancer (likely due to easier, better diagnosis).
ReplyDeleteThese trends are real, though as you alluded to in your post, the cause isn't always evident.
I have been impressed by the large number of patients with CLL who have been born between 1940 and 1965. Although I never practiced hematology, oncology I must say that in 27+ years of practice I can't recall ever seeing any patients in their 50's with the diagnosis of CLL, yet now that I have it in my 50's the world seems awash in similar patients. The question is, are we diagnosing it earlier, or has there been a fundamental change in the disease. Perhaps there is something akin to a new cohort of patients with CLL, diagnosed in their 40's, 50's and early 60's with a more aggressive course of disease.
Only time, study and careful follow up will tell, though I'd be curious to hear an expanded view of your take on this.
DWCLL
The question I have to ask is, how relevant is this study of patients in predicting outcomes for recently-diagnosed patients today?
ReplyDeleteI mean, were patients getting the same treatment during the 30 years prior to 2004? Isn't, say, Rituxan relatively new? And many other things?
Well, of course.These data only go up to 2004. However, fludarabine was widely used in the previous decade and FC, FR and FCR were both used increasingly. Clearly, these patients might not yet had an impact on 10-year survivals, though they may have had an effect on 5-year survivals.
ReplyDeleteI certainly hope that we are seeing an improved survival now. But the purpose of my article was to warn about being taken in by the statistics. There are very good reasons why an improvement in survival might apparently occur without there being any real change.
I like to think of this a line with the start point = diagnosis and the end point = death. The study in last month's Blood is clear that the line has grown longer over time - patients are living longer after diagnosis. So patients who are newly diagnosed are at the start point, and they can take heart during those initial shocking days.
ReplyDeleteBut it is unclear if the line is longer because either or both end points have moved. It is also unclear about increased survival at any point along the line other than diagnosis - you cannot tell where along the line you are except at initial diagnosis. You have to be able to put your finger on where you are at along the line, and the only reference point is initial diagnosis.