Sunday, December 2, 2012

More on the Issue of Screening

The problem of cancer screening has been at the forefront for several years. The two most visible are that for prostate and breast cancers. Let me make a few observations and then I want to use the recent NEJMdiscussion on breast cancer as a discussion point[1].

Let us examine a single cancer, be it breast or prostate, or frankly almost any other. Let us look at steps.

1. At some time unbeknownst to the patient and physician some genetic pathway aberration occurs. It may be a transposition such as in CML or BRAF V600 as in melanoma or PTEN as in a prostate cancer. This is time 0. Something happens which sets off a chain of subsequent events.


2. Let us assume that the patient can be doing one of four things:

2.1 The patient is, with his physician’s guidance, undergoing genetic surveillance. Namely we assume that there is some genetic marker which has high specificity for detecting the remnants of this genetic change. We do not know what it is yet but it gets the cancer when say there are a few thousand cells. It has a high detection probability and a very low false alarm rate. Namely, it is a good test. It may or may not determine if the cancer is aggressive or indolent. Namely it has undetermined prognostic value. The patient if found to have a positive result will undergo a detailed invasive procedure to affirm the diagnosis and then if that is positive then undergo normally accepted surgical and chemotherapeutic treatment.

2.2 The patient is, with his physician’s guidance, undergoing some form of marker testing for cancer surveillance. This may be a PSA or mammogram test. It may have a lower detection probability and a higher false alarm rate.  As with the first scenario the patient will undergo further examination. The patient if found to have a positive result will undergo a detailed invasive procedure to affirm the diagnosis and then if that is positive then undergo normally accepted surgical and chemotherapeutic treatment.

2.3 The patient, with his physician’s guidance, will from time to time be clinically examined to determine if there is an abnormality. Thus for prostate exams it is a DRE and for breast exams it is a palpating of the breast. If no palpable lesion is present no other tests will be performed. If a palpable lesion is present then normal next steps performed. The patient if found to have a positive result will undergo a detailed invasive procedure to affirm the diagnosis and then if that is positive then undergo normally accepted surgical and chemotherapeutic treatment.

2.4 The patient will see their physician only after they have determined that a problem exists. Namely, in the case of the prostate, back pain is present along with the inability to urinate. For a skin lesion, the pigmented lesion is raised and bleeding. For the breast, the lump is obvious to sight not just touch. This is the way things were a century ago. The patient will then present to the physician. The patient if found to have a positive result will undergo a detailed invasive procedure to affirm the diagnosis and then if that is positive then undergo normally accepted surgical and chemotherapeutic treatment.

3. Treatment is initiated at the time we have specified above. We know the age dependent incidence for all such cancers so we can determine the incidence readily. Now for the outcomes.

3.1 If a genetic test determines early cancers, then local excision may be effective allowing for low costs removal. Follow up care be accomplished as well with genetic follow up. The death rate therefore should be near zero except for those high degree metastasis cancers such as ovarian, glioma, and certain melanomas, for example.

3.2 If we have intermediate tests whereby we can reduce chances of metastasis and excise in early cases then we can reduce the disease related mortality by a material degree.

3.3 If we wait for clinical presentation albeit with periodic surveillance then we may have a larger degree of metastasis and thus substantially higher mortality.

3.4 If we wait for the patient to inform the physician then most likely we have the highest metastasis and highest mortality.

Data relating to the above specific mortality can be derived based on the number of cancer cells at point of detection. For example for genetic markers of PCa we may be able to accomplish this with a tumor load of 1,000 cells. For PSA we need most likely 100,000 cells and for palpable screening based tumors we need 10,000,000 or more cells[2].

4. What we should measure is death from the specific cancer, not death. But that is also complex. For example if a 75 year old man with congestive heart failure is determined to have PCa, then he most likely will die of the cardiac problem than the PCa. Yet if we have death as the all-encompassing end point the results will not be conclusive.

Let us now examine the NEJM comments.

1. No Treatment or Screening for Anyone: The respondent argues:

More than 600,000 women were enrolled in several randomized trials of mammography screening more than 30 years ago. The trials showed a reduction in breast-cancer mortality but no effect on all-cause mortality. In recent decades, breast-cancer treatment has improved greatly, resulting in reduced mortality even among women with advanced disease. Thus, the reduction in mortality from mammography screening may be significantly smaller today than when the trials were performed.

The statement is somewhat conflicting in my opinion. If a reduction in mortality from breast cancer was observed then that is good. Even if we have improved treatment on later stage disease that is in my opinion no reason to delay early treatment.

The conclusion is:

The decision regarding screening mammography depends on the balance of benefits and harms and the way in which a woman and her physician weigh these competing factors. In my view, the benefits do not exceed the harms. Thus, given the data currently available, I do not recommend mammography screening for average-risk women of any age. I would ask the woman in the vignette to come back when she is 50 years of age. In 10 years, we might have more data to provide a different recommendation for her.

Here we have in my opinion a conflict and an observation. Bayesian analysis always comes to bear. If we have a family history then we had better look more closely. We know that such a history raises risks. But the problem often is uncertain family history as well. How do we handle that/

2. Treatment and Screening at 40

At the extreme the respondent argues:

Screening can be thought of as a kind of insurance. As with all insurance, there are costs for protection against adverse events that have a low probability of occurrence but could be catastrophic if they occurred without the insurance. In that context, given the evidence, there are good reasons to begin screening at the age of 40.

I would agree with this argument. It does save lives, albeit less than economically worthwhile to some, and yet it provides a data base for further analysis as well. The morbidity from false positives is always a concern but frankly that is a patient’s choice as it always should be.

Thus for a conclusion, in my opinion it is clear that the logic for no screening hsa little if any merit, that for early and continual screening with informed consent, and even with an added payment, should be the norm.


[2] See Weinberg, Cancer, Garland, 2008. pp 361-364, and the actual number of such cells may vary dramatically by cancer type and the issue of cancer stem cells are critical as well.