Monday, January 27, 2014

Prostate Cancer Overdiagnosis?



Prostate Cancer is the number one occurring cancer in men in the U.S. At the same time there is an ongoing debate as to the need to screen for this cancer. The problem is that most cases of PCa are indolent and will not be the cause for a man’s demise. On the other hand we know that a significant number of cases, 5-15% of them, are aggressive and will result in a very painful death in a short period, two to six years or less. The problem is that there is no gold standard test to determine which is which.

Various genetic profiles have been proposed wherein they measure the expression of a panel of genes and then calculate a metric, usually some number, which if in a certain range means indolent and outside of the range is aggressive. The problem of course is; what cells are you making this measurement on? If you are doing it on the encapsulated prostate cells then you may be missing the already metastasized cells which have moved to the bone.

Now a recent paper by Gulati et al state[1]:

The chance that a prostate cancer detected by screening is overdiagnosed (ie, it would not have been detected in the absence of screening) can vary widely depending on the patient’s age and tumor characteristics. The purpose of this study is to use age, Gleason score, and prostate-specific antigen (PSA) level to help inform patients with screen-detected prostate cancers about the chances their cancers were overdiagnosed.

First I would be concerned with the definition of overdiagnosed. It states; “it would not have been detected in the absence of screening”. One should examine this. If one screens and detects a PCa then that is an overdiagnosis. If, however, one gets a patient who comes into your clinic with massive back pain and dysuria, then that patient is not overdiagnosed. The latter patient is however terminal. Thus I would strongly quibble with this definition of overdiagnosed.

They continue:

A computer microsimulation model of prostate cancer natural history was used to generate virtual  life histories in the presence and absence of PSA screening, including an indicator of whether screen detected cancers are overdiagnosed. A logistic regression model was fit to nonmetastatic patients diagnosed by screening with PSA less than 10ng/mL, and a nomogram was created to predict the individualized risk of overdiagnosis given age, Gleason score, and PSA at diagnosis. The calibrated microsimulation model closely reproduces observed incidence trends in the Surveillance, Epidemiology, and End Results registries by age, stage, and Gleason score. The fitted logistic regression predicts risks of overdiagnosis among PSA-detected patients with an area under the curve of 0.75. Chances of overdiagnosis range from 2.9% to 88.1%. The chances of overdiagnosis vary considerably by age, Gleason score, and PSA at diagnosis. The overdiagnosis nomogram presents tailored estimates of these risks based on patient and tumor information known at diagnosis and can be used to inform decisions about treating PSA-detected prostate cancers.

Now I would also have concern as regards to such a model. We have been examining them in several cancers and they are complex and require data which we are yet able to provide. Moreover a logistic analysis is rant with many problems; it merely hypothesizes a relationship based on a correlation model which may bear no resemblance to reality. In its place one really needs a tempero-spatial model which includes genetic mutations in some Markov manner. Also I would be concerned with an overdiagnosis range of from 2.9% to 88.1%.

Now a comment by Rathner in the same issue states as follows[2]:

Using a nomogram that incorporates age, Gleason score, and prostate-specific antigen (PSA) level at diagnosis, individual risks that a screen-detected prostate cancer has been overdiagnosed can be estimated, according to a new study published January 6 in the Journal of the National Cancer Institute . The authors used a standard definition of overdiagnosis to refer to a cancer that would not have become symptomatic or clinically identifiable if it had not been detected by screening. Overdiagnosed cancers do not pose a risk to the patient and do not require treatment, which is associated with significant risks of impotence and incontinence.

Here there is a clarification of overdiagnosed as meaning indolent. Indolent means slow growing and of de minimis risk of death from the lesion. However it is highly problematic to make such a determination unless one uses genetic metrics on a whole body basis. Techniques using exosomes may be beneficial if the profiles are stable.

Previous studies have estimated the risk of overdiagnosis for the U.S. population, with results ranging from 23% to 42% of screen detections. However, risks of overdiagnosis can vary considerably depending on the patient’s age and tumor characteristics, highlighting the need for a personalized tool to predict the likelihood of overdiagnosis. … The authors used a microsimulation model to generate virtual life histories for a representative population of u S men between 1975 and 2005. Men who develop cancer can be detected based on elevated PSA levels or development of symptoms. ,,,,

 A prediction model was then developed to predict individual chances of overdiagnosis (i.e., the chance that other-cause death would precede diagnosis in the absence of PSA screening) given information known at screen detection. The prediction model estimates that the chances of overdiagnosis range from 2.9% to 88.1% depending on patient age, PSA, and Gleason score. …

Freidlin and Korn question whether the Gulati et al model of the risk of overdiagnosis is useful in guiding treatment decisions of patients with screen-detected prostate cancer: “...once an individual has been screened and found to have prostate cancer, the relevant question is the outcomes of various treatments (treatment morbidity, prostate cancer symptoms and death), and not the probability of an event [detection of prostate cancer] that could have happened if the individual had not been screened.”

One then must ask if it is ethical to perform a fully blinded randomized trial to ascertain the predictions made herein. As much as I am a fan of models, this is not a phenomenological model. It merely uses data from previous diagnostic tests to ascertain the importance of certain metrics. As such it lack what I believe is a sine qua non to approaches like this, a physical model with predictability.

Now comments in Healio state the following[3]:

Researchers assessed risk for overdiagnosis — defined as a cancer detected through screening that otherwise would have been asymptomatic or clinically unapparent— using a microsimulation model of virtual life histories of men aged 50 to 84 years from 1975 to 2005.
Researchers then applied SEER prostate cancer incidence and PSA screening data to the virtual models. Results indicated that the odds for overdiagnosis increased by 12.9% (95% CI, 12.2-13.6) for each additional year of age at the time of diagnosis.

One of the concerns is that using SEER from 1975 to 2005 may insert a bias in age since PSA was not used to any extent until 1995 at the earliest. Thus half the data was without PSA testing and thus the older men may very well already have PCa.

A Gleason score of at least 7 was associated with a 19.5% (95% CI, 11.7-26.5) decrease in the risk for overdiagnosis when compared with a Gleason score of 6 or lower (P<.001). The odds of overdiagnosis decreased by 16.6% (95% CI, 14.2-18.9) with each additional 1 ng/mL of serum PSA up to 10 ng/mL.

A Gleason of 7 on biopsy may be really a Gleason of 8 upon prostatectomy. It may even be higher. PSA is also an issue related to age, prostate volume, BPH, and the better measurements are those reflecting temporal changes. Single PSA measures have the same problem as spot blood pressure measures. Researchers found age to be the most statistically significant risk. Among men with a Gleason score of 6 or lower and PSA levels from 4 ng/mL to 4.9 ng/mL, those who were aged 50 to 54 years had an 11.6% risk for overdiagnosis, whereas those who were aged 70 to 74 years had a 59.9% risk and those who were aged 80 to 84 years had an 83.4% risk.

Age has always been a significant factor and the results presented here in my opinion just reiterate that fact.

The question is; does such a nomogram have any clinical value? That is in my opinion problematic. The handful or so of first line Urologists at major centers will have thousands of cases where they can ascertain based on a plethora of data a patient’s prognosis, yet even they are wont to go too far. Patients, in my opinion, may find such an approach as just another element to add to an already confusing pile of suggestions.

References

Gulati R et al, Individualized Estimates of Overdiagnosis in Screen-Detected Prostate Cancer, JNCI J Natl Cancer Inst (2014) doi: 10.1093/jnci/djt367 First published online: January 7, 2014

Rathner, Z., Nomogram to determine individualized estimates of screen-detected prostate cancer overdiagnosis, JNCI J Natl Cancer Inst (2014) doi: 10.1093/jnci/dju001 First published online: January 7, 2014