Our imaging study findings in 17 patients support the dual-ventral-bud embryological theory of annular pancreas put forward by Baldwin[9]. The presence of two ventral buds would result in two major pancreatic ducts, in the pancreatic head and annulus, respectively, according to one of the three configurations described in Lin’s classification. All 17 patients had ductal configurations matching Lin’s description.
Many theories have been suggested to explain the development of an annular pancreas. The normal pancreas originates in a dorsal and a ventral bud, which arise from the primitive foregut between the 5th and 7th weeks of gestation. The buds rotate and eventually fuse into a single gland. Lecco suggested that an annular pancreas forms when the free end of a single ventral bud adheres to the duodenal wall, is pulled into a ring shape when the duodenum rotates, and finally fuses with the dorsal bud. This theory explains neither the presence of two ducts from the ventral bud (annular and left ventral) nor the existence of non-divisum forms of annular pancreas, as the ventral duct would be unable to fuse with the dorsal duct if its free end were directed anteriorly instead of posteriorly. Baldwin’s hypothesis assumes that the ventral bud has two lobes. Normally, involution of the left lobe would precede the rotation of the pancreatic anlagen. Baldwin suggested that the left lobe might persist, adhering at its tip to the duodenum. In this case, the left lobe and its duct would be pulled around the duodenum during its rotation. As described in Lin’s classification[10], the free end of the right ventral duct would then be able either to fuse with the dorsal duct, forming the main duct of Wirsung (type II or branch annular pancreas), or to remain independent, leading to the divisum configuration (type I or divisional annular pancreas). In some rare divisum cases where the right ventral and dorsal ducts fail to join, the free end of the annular (or left ventral) duct would fuse anteriorly with the dorsal duct, resulting in a main pancreatic duct encircling the duodenum (type III or main duct annular pancreas).
Several pathological and embryological studies in animals support Baldwin’s hypothesis: a smaller left ventral bud that usually regresses has been reported in mammals[14], and a paired ventral pancreas exists in fully formed non-mammalian animals such as chicks and Xenopus[15, 16]. The only data for humans, however, come from the postmortem examination of a baby who died 1 day after birth at 32 weeks of gestation, which showed an annular pancreas originating from the left lobe of a bilobed ventral bud[17]. It was therefore of interest to assess ductal configurations in patients, as an effort to determine whether they supported a particular embryological theory.
We identified three main ductal configurations, which matched the three types described by Lin. Type II or branch annular pancreas was by far the most common type, with nearly two-thirds of cases, in line with previous studies[1, 11, 18, 19]. Type I or divisional annular pancreas contributed about two-fifths of cases. Our results confirm that pancreas divisum is more common in patients with than without an annular pancreas (4–10%)[20], in accordance with the previously reported rates of 29%[3], 37%[1] and 45.7%[11]. Relative atrophy of the right ventral bud in contrast to the hypertrophy of the left ventral bud may explain the higher frequency of divisional forms[10]. Type III was found in only one of our patients, implying that it is the least common type[10, 21].
We excluded patients who had an incomplete parenchymal annulus. Incomplete annular pancreas was found in 37.5% of patients in a retrospective database study examined by ERCP and/or MR imaging and/or CT[1]. This proportion may have been overestimated, however, as the CT slice thickness was 5 or 6.5 mm, precluding multiplanar reconstruction, which plays a crucial role in detecting rings that are thin, obliquely-oriented, and/or hypodense due to fat-related changes. A crocodile-jaw appearance of the pancreatic head with no evidence of a complete annulus by multiplanar reconstruction or of an annular duct by MRCP should be considered a head-contour variant rather than an incomplete annular pancreas. We also excluded patients with circumportal annular pancreas as the origin of this variant does not involve a supernumerary pancreatic bud but is believed to originate from a hypertrophic uncinate process[22].
To our knowledge, ours is the largest study to date in which the ductal configuration of annular pancreas was assessed using noninvasive cross-sectional imaging studies. Our results confirm that annular pancreas is rare, with only 17 cases identified over 12 years at our large institution. However, annular pancreas often escapes diagnosis. Half our patients presented with acute abdominal pain, a situation in which radiologists specialized in the gastrointestinal tract are more likely to read the imaging studies and to detect annular pancreas if present. Thus, the proportion of cases of annular pancreas with abdominal pain in our study is probably an overestimation. The marked predominance of females in our study is consistent with earlier data[1, 5].
The presence of an annular pancreas was clinically significant in three patients, modifying the treatment strategy in one and resulting in additional diagnostic investigations in the other two. Although annular pancreas is usually an incidental and asymptomatic finding, the ductal configuration is clinically significant in patients undergoing ERCP or pancreaticoduodenectomy. Thus, in Type III, incision of the pancreatic neck may result in a leak from the main annular duct.
The main limitation of our study is the retrospective design. We used keywords to identify potentially eligible patients in our database, as reviewing all imaging studies performed over the 12-year period would not have been feasible. Whereas ERCP was often used in earlier studies, we relied on CT and/or MRI and MRCP to describe the ductal configurations. ERCP is now indicated only for therapeutic purposes[23].
Our sample size was small, and we cannot rule out that some forms of annular pancreas not represented in our population fail to support Baldwin’s theory. Thus, the rare cases of annular duct draining directly into the minor papilla cannot be explained by the dual ventral bud theory.