Probably every lung doc in his career has seen a patient who has been apparently cured ( at least has gone over five years without evidence of disease) after a lung cancer was incidentally detected and then resected . The typical scenario was that an incidental chest xray demonstrated a solitary pulmonary nodule (SNP). ( A single nodule less than 3 cm in diameter) With the prognosis of lung cancer detected at a symptomatic stage so dismal it is understandable that we would yearn for some method of early detection.There is no general agreement that such a method has been found and some wonder if it is even possible.
The clinical trials that examined results of screening with annual chest x-rays have not clearly shown benefit and arguably suggested harm. This has resulted in most groups not recommending periodic chest film screening even in smokers.
Lung CTs are the latest candidate for a early detector of lung cancer and that technique has not been warmly received even though pulmonary physicians certainly would welcome something to make the prognosis of detected lung cancer much better as progress in that area is not one of the cancer cure success stories of modern medicine.
A number of articles have been published in which the authors concluded that CTs are very helpful and have recommended their application. Now in the WSJ health blog we are given some details of what seems like fairly big red flag involving conflicts of interests and failure to disclose those potential conflicts in some of these researchers who have written and spoken widely advocating CT screening.
One of the major "pro-screening" articles was published in the NEJM and I have commented about that before as well as delving into the issue of "how do we decide if screening really works ?" I think much of that answer depends on if you think survival rates ( e.g five year survival rates) or cause specific mortality for a specific cancer (I am note sure if we can ask that about cancers in general as I believe it depends on the natural history of the cancer) should be the statistical arbiter. Most people who write about this favor cause specific mortality and that is the conventional wisdom but there is at least one articulate advocate of the other view mentioned in my above self reference. Which view I take seems to depend on whose article I have read most recently.
The contrarian in this regard is Dr. Gay M. Strauss from Dana-Farber Cancer Institute.He presents his case in Chest 1997:112 216 s-228s ) "Measuring Effectiveness in Lung cancer Screening." I think a subscription is required. Anyone interested in this issue will find his argument either interesting but flawed or convincing and spot on or will be unable to decide which assessment is correct.
His argument suggesting that RCTs may not be the final answer regarding the efficacy of screening for various cancers does not convince me because I really don't understand it. Whose fault is that? He asserts that small absolute differences in disease risk between screened and control group often persist after randomization and translate in large proportional difference in the size of sub groups at risk for disease specific mortality and may give misleading results when a randomized clinical trial (RCT) is applied in the preventive medicine arena. He is not arguing against the value of RCTs in other contexts.
One of his other arguments resonates with me to a much greater degree. One anti-screening argument is that survival rates look favorable for screening while mortality rates do not is explained by the detection of large numbers of indolent lung cancers which really did not need to be treated. This argument is commonly made in regard to prostate cancer. Strauss argues that this is not valid as regards lung cancer. Now if this "lead time bias" is at work here, he continues,therapy would irrelevant to outcome. So let us look at the outcome of those stage 1,non-small cell cancer patients who were screened but did not undergo surgical resection .What is their outlook. Five year survival was 5%- rather low survival for supposed indolent cancers. This is far less than the what we see with resected lung cancers that present clinically as SNPs.
Prostate cancer is not rare in autopsy series-some have shown over 50% prevalence of latent prostate cancer so the concept of indolent or very slow growing prostate cancer is very reasonable while "surprise" lung cancer at autopsy , according to Strauss, has a prevalence of 0.8%. Again if we are detecting this not clinically important entity by CTs , why do we not see large numbers of lung cancers at autopsy? ( It should be noted that at least one autopsy series showed a much higher number of previously undetected lung cancers.)
So who is right? I don't have the answer but I believe that there is more to it than simply closing the debate by shouting firmly " Lead time bias ".
Now entering the discourse is the accusation of possible conflict of interest in some of those who are recommending lung CT screening.In a world where decisions were made with the pure logic of a Mr. Spock the possibility that someone taking a particular position possibly gaining from acceptance of that position should not be determinative in the debate about the truth of that argument, but in the non Spock world thorough we pass people who have no Vulcan genes definitely take into account the "credibility of the witnesses". ( In some settings Spock would consider that also).
As pointed out by Dr. Howard Brody, conflict of interest is not localized to Big Pharma and
"Nothing like knocking public trust in academic medicine and medical research down a few more notches."