| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Shands-Jacksonville Hospital Jacksonville, FL 32209
We read with interest the article entitled "CT of primary Cystic Pancreatic Neoplasms: Can CT Be Used for Patient Triage and Treatment?" by Curry et al. [1]. After reading the abstract, the senior author of this letter (21 years of CT experience) looked at all images with the captions covered and tried to make a diagnosis. He then recruited the two other authors (14 and 7 years of CT experience) to do the same. All authors diagnosed the same six of the nine cases correctly. Two of the remaining three diagnoses were judged inconclusive because the masses were so small. The sole error was in Figure 5, depicting the large round, smooth pancreatic tail mass with no calcification or internal architecture. All authors thought this was a mucinous cystic neoplasm when it was, in fact, a microcystic adenoma. This case may indeed represent an "atypical serous cystadenoma." Alternatively, the contrast enhancement was not optimal, and a better scan or another imaging modality might have revealed fine internal septations. What was the gross pathologic correlation?
As pointed out by Curry et al., [1] the error of calling a microcystic adenoma a mucinous cystic neoplasm is less serious than the reverse. Our accuracy seems higher than that reported by Curry et al. Perhaps their reviewers' accuracy would have been higher had the study included a category of "too small to characterize." Lesions in this category are probably incidentalomas and probably should be followed up. We do not know if the illustrated cases are representative of the material as a whole, but using this strategy we would have been correct in six of seven diagnoses.
The CT criteria used by Curry et al. [1] consisted of dimension of the largest cyst, type of calcification (if present), metastatic disease, and adenopathy. They did not make use of previously published criteria such as papillary excrescences or solid masses (mucinous cystic neoplasm); external lobulation (microcystic adenoma); external smoothness (mucinous cystic neoplasm); swiss-cheese contrast enhancement (microcystic adenoma); and lack of enhancement except in septa, excrescences, or wall (mucinous cystic neoplasm) [2, 3]. Perhaps these additional criteria, if used, would have improved accuracy.
A minor quibble is with the use of the word "insensitive" in the sentences in the conclusion section of the abstract and the final sentence of the article, which describe CT as "an insensitive tool for differentiating these tumors." Surely the authors meant either "inaccurate" or "nonspecific." Finally, the statement, "Despite more than 30 years of combined experience interpreting CT examinations, our reviewers were uncertain of the diagnosis in an average of 57% of patients" may be misleading. Two of the authors probably have more than 20 years of experience, but we don't know if either were among the three retrospective reviewers.
In conclusion, we feel more optimistic about the clinical usefulness of CT evaluation of cystic pancreatic masses than do Curry et al. [1].
References
Johns Hopkins University School of Medicine Baltimore, MD 21287
My coauthors and I appreciate the comments of Friedman et al. on our article [1], which addressed the ability of three reviewers to definitively distinguish between serous cystadenomas of the pancreas and mucinous cystadenomas or cystadenocarcinomas of the pancreas. Our goal was simple and at the same time practical in terms of daily clinical practice. As CT use increases, we are seeing more incidental pancreatic masses and more cystic pancreatic lesions, especially in older patients. Although numerous articles describe the classic CT appearance of the various cystic pancreatic tumors, to our knowledge little detailed analysis of the success of these criteria has been published. Our study reviewed the cases of 50 patients, gathered from a 14-year period, who had pathologically proven tumors. Scan protocols and scanner quality obviously varied over the time period, but we did not attempt to adjust for this variable. Rather, we decided to look at all cases as they would be reviewed in a true clinical environment, especially in a tertiary medical center.
As noted in the article, the reviewers had less than optimal success in distinguishing among the various tumors. In fact, although I was not one of the three reviewers, I too was surprised by the results of the study. Those of us who interpret body CT on a daily basis often see or are shown cystic pancreatic lesions that we assume are benign, and we recommend little more than conservative follow-up with repeated CT at a later date. We do this without pathologic proof and in all likelihood are correct in most cases. Yet our true accuracy goes undefined, because most of these patients never undergo biopsy or have surgery. This point is one of the take-away messages of the article: We may be overestimating our own accuracy in identifying the various cystic pancreatic lesions.
We now have available multidetector CT using dual-phase imaging and routine scan collimation in the 1- to 2-mm range, as well as CT angiography. I expect our success rate will increase as we continue to obtain better definition and detail of a pancreatic mass. Such has been the case in our ability to identify other pancreatic lesions, like islet cell tumors, and in staging pancreatic cancer. Perhaps in the near future we will see in the literature a repeat of our study that will consider only dual-phase multidetector CT cases.
Interestingly, within a few months of our article Procacci et al. [2] published similar work, which looked at 100 patients with cystic tumors of the pancreas. These authors found results similar to ours and noted that distinguishing between serous and cystic pancreatic tumors is difficult when using CT criteria alone. Our reported experience, therefore, does not appear to be an isolated event, but may be truly representative of radiology's abilities in this difficult arena.
References
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
