Fax: (786) 596-5986
Reporting risk of malignancy/dysplasia in cytology
A potential way to improve communication, if not reputation
Article first published online: 9 OCT 2007
Copyright © 2007 American Cancer Society
Volume 111, Issue 6, pages 465–466, 25 December 2007
How to Cite
Renshaw, A. A. (2007), Reporting risk of malignancy/dysplasia in cytology. Cancer, 111: 465–466. doi: 10.1002/cncr.23123
- Issue published online: 17 DEC 2007
- Article first published online: 9 OCT 2007
- Manuscript Accepted: 4 SEP 2007
- Manuscript Revised: 29 AUG 2007
- Manuscript Received: 20 AUG 2007
Cytology can be used in 2 different ways: as a diagnostic test or as a screening test. As a screening test, a cytologic diagnosis is associated with specific levels of risk for malignancy or dysplasia. There are many different words for communicating this risk. These include the well-known negative, atypical, suspicious, and positive diagnoses. However, there are also many more subtle words and phrases: suggestive of, suspicious for, cannot rule out, and consistent with, just to name a few. To every cytologist these different phrases have very specific meanings and levels of risk, and in our department we have had many discussions regarding which of these phrases is best for an individual case. Nevertheless, many of our clinicians do not have such a sophisticated approach to these cases. Either the cytologist can make a diagnosis or he cannot; either he knows the answer or he does not. For these clinicians, all cytologic tests are diagnostic tests, and the idea that some interpretations measure risk very accurately is unknown. At the very least, this represents a communication problem. How can we improve this situation?
One approach would be to try and educate our clinicians concerning the various phrases in use. However, this would be labor intensive and most likely not well received by our busy clinicians, and it is my impression that even cytologists disagree about exactly what level of risk each different phrase represents. In addition, in some settings, the phrases used are simply not very good at describing the risk of an individual case because they are not consistent and may even be contradictory. For example, in thyroid aspirations specimens, a case that is worrisome for a Hurthle cell neoplasm is often diagnosed as being suspicious for a Hurthle cell neoplasm (or even positive in some centers). Yet even with the worst outcome, the risk of malignancy is reported to be no higher than 20%.1 In contrast, a case with just a few clusters of cells with features of papillary carcinoma is often signed out with no worse a label than atypical cells present, a papillary carcinoma cannot be ruled out. Nevertheless, this case is associated with a risk of malignancy of no less than 30%.1 As a result, it appears unlikely that any amount of education on the part of the clinician is going to clarify terminology that simply is not used consistently by cytologists.
A better approach, in my opinion, would be to stop emphasizing these different phrases and instead report quantitative assessments ofrisk in cytology. Every cytologist knows there is a big difference in meaning between a case that is highly suspicious and one that has a few atypical cells of unclear significance. The suspicious case has a very high level of risk that most likely is quantifiable based on either the literature or the laboratory's and the individual cytologist's past experience. The atypical case most likely has a very low level of risk of malignancy, which may be very difficult to characterize. If we reported that 1 case had a risk of malignancy of approximately 80% to 90% and another case had a low but undefined level of risk, at the very least we would help to distinguish between indeterminate cases that have a well-defined risk level and those that do not.
Some cytologists might say that it is not possible to measure the risk of malignancy in any particular case because the risk in an individual case depends on numerous factors, including clinical factors that may not be known to the cytologist, and the diagnosis that the cytologist makes is subjective and not necessarily reproducible. Although these facts are true, the conclusion is not, and represents a misunderstanding of both statistics and the cytology literature. For example, when one buys a lottery ticket, the actual chance of winning is either 0% or 100%, and depends on a variety of factors that the buyer cannot know including (for cases drawn from a hat) where the ticket is in the hat, and where the person drawing the ticket puts his hand. However, this does not mean that a savvy statistician cannot estimate his chance of winning quite accurately based on the number of tickets in the hat. Indeed, much of the literature consists of examples of just this sort of risk estimation, including the risk of type 3 cervical intraepithelial neoplasia when given a diagnosis of atypical squamous cells of undetermined significance (ASCUS) on a Papanicolaou smear2 or the risk of malignancy for a suspicious diagnosis based on an endoscopic ultrasound -guided fine-needle aspiration of the pancreas.3
Of course, reporting quantitative risks can be frightening because it makes the cytologist report exactly what he means. This might make some cytologists very uncomfortable. There are several ways to alleviate this discomfort. For example, in many cases, a range of risk may be the most appropriate way to express the risk, rather than a single specific number. In addition, it might be worth reporting the risk ranges from the literature, the individual laboratory, and the individual cytologist, all of which may be very similar or very different depending on how individual cytologists practice and their patient population. One could even consider presenting the data on which the laboratory or individual cytologist is basing the risk assessment. This would give a very different meaning to personalized medicine. In this way, the clinician can chose exactly which risk assessment he prefers to use. If he is very familiar with and likes the work of a particular cytologist, he can use the individual risk as his guideline. If he is not familiar with a particular cytologist, he can use the individual risk as a way to compare the assessment of that cytologist with those of others. An added benefit of this method would be that the cytologists who review the most cases and have the most follow-up and, presumably, are the most reliable diagnosticians would be able to document in the report exactly how experienced they are. This experience could be used by the clinician to help in determining the value of their specific risk assessments.
If this approach were adopted, one could expect to have 4 different parts to every cytology report: a statement of adequacy, a diagnosis, a risk level (or 2 or 3 or “risk of malignancy uncertain”), and the data on which the risk level is based. The system might be applicable to gynecologic as well as nongynecologic cytology. After all, although ASCUS is not associated with a specific histopathologic diagnosis, and may be characterized by some clinicians as a diagnosis without a specific meaning, it is associated with a very specific and known level of risk of dysplasia that can be reported and used in making clinical decisions.2 In those cases in which risk assessment is appropriate, this approach might improve our communication with clinicians, and may help convince some clinicians that not every indeterminate report is ambiguous and therefore worthless. In my opinion, it is time for the cytology community and the societies that represent it to demand that cytologists stop hiding very specific, accurate, and clinically useful risk assessment information behind diagnostic terminology that is ambiguous to everyone except the individual cytologist who uses it.