Hindsight Is Not 20/20: a Cautionary Tale in Pancreatic Cancer

It is widely recognized that endoscopic ultrasonography (EUS) in combination with fine-needle aspiration (FNA) has a specificity of almost 100 % in the diagnosis of pancreatic cancer. While EUS is in our view a very sensitive diagnostic tool, the diagnosis of pancreatic cancer can be overlooked even by expert endosonographers. Short of direct procedure-related complications, missed gastrointestinal cancers are the most dreaded and most humbling consequence to face the practicing endoscopist. In a study published in this issue of Endoscopy, Bhutani et al. [1] sought to identify factors leading to a missed or delayed diagnosis of pancreatic cancer on EUS. While this cautionary study identified practical limitations of EUS-FNA in detecting pancreatic cancer, there are several important caveats regarding it which need to be addressed.

The principal shortcoming of the report lies in its retrospective nature, with the resulting intrinsic recall bias. At busy academic centers in which endosonographers often perform over 40 examinations per week, recall of individual case details is difficult at best. While it is not clear how much time passed between the index EUS examination and the questionnaire response, we suspect there was a delay. Furthermore, selection of cases was dependent on each endoscopist’s own clinical database search. At referral centers in which patients may be evaluated only once, follow-up is never complete (the authors recognized this), and relying on self-reporting of errors presents an additional opportunity for selection bias [2]. These points question the internal validity of the study.

Although many pancreatic cancer EUS examinations are indicated on the basis of abnormal abdominal computed tomography or magnetic resonance imaging, it is becoming more common to use EUS for the evaluation of ”dilated ducts, unknown cause.” In this survey, all of the patients included had undergone other negative imaging examinations beforehand (with the exception of the 20 % who had a dilated ductal system), of unspecified quality. This suggests that these patients had very early lesions (which may or may not have accounted for their symptoms) that simply lie beyond the limits of any combination of diagnostic approaches. It is not clear how long the false-negative EUS delayed the diagnosis, or what ultimately prompted surgery. The fact that this group of patients was highly selected, and that only very experienced endosonographers were queried, further limits the ability of the report’s findings to be generalized.

These comments notwithstanding, the report identifies several common pitfalls, of which all endosonographers should take note. A prominent ventral-dorsal anlage, diffusely infiltrating carcinoma, incompletely resolved acute pancreatitis, coexisting chronic pancreatitis, as well as stent artifacts and a location near the uncinate process or genu are all features that worry us in the hunt for pancreatic cancer. Fortunately, EUS is continuing to evolve, and among the more promising new diagnostic techniques is using a combination of FNA and gene expression profiles to achieve more accurate detection of cancer at an earlier stage. Such an approach would be particularly useful in patients in whom pancreatic cancer coexists with and is confounded by an intense stromal reaction (chronic focal pancreatitis).

The optimal study design for identifying pitfalls in the EUS diagnosis of pancreatic cancer would be a multicenter one (including smaller community EUS centers), with all patients referred for a suspicion of pancreatic cancer being followed up prospectively. All examinations would need to be recorded at full length for subsequent review, and the questionnaire would have to answered immediately after the initial examination. It would be important to know what additional factors confound the diagnosis of pancreatic cancer for examiners with less than an expert level of experience.

If the suspicion of carcinoma remains high despite negative imaging, we opt for a repeat EUS within 4 weeks, and we may suggest that another physician in our group perform the follow-up examination. In cases in which a discrete mass is not seen but the duct is observed to taper abruptly, it may be worth performing empirical biopsies to detect infiltrating disease, although this is not standard practice. We suspect missed cancers are more prevalent than we like to think, and a more rigorous, prospective trial would be best way of clarifying this very important clinical situation.

References

• 1 Bhutani M S, Gress F G, Giovannini M. et al . The No Endosonographic Detection of Tumor (NEST) Study: a case series of pancreatic cancers missed on endoscopic ultrasonography.  Endoscopy. 2004;  36 385-389
• 2 Newcomer M K. Underestimation of adverse events following ERCP: a prospective 30-day follow-up study [abstract].  Gastrointest Endosc. 1995;  41 408