Quantitative determination of pancreas size using anatomical landmarks and its clinical relevance: A systematic literature review

INTRODUCTION

Diseases of the endocrine and exocrine pancreas entail an increasing health and socioeconomic burden globally. Diabetes mellitus, a spectrum of chronic metabolic disorders characterized by chronic hyperglycemia with a central component of pancreatic endocrine dysfunction, has been reported to affect >400 million individuals worldwide. In 2015, 1.6 million deaths were directly attributable to diabetes mellitus, and hyperglycemia was a significant risk factor in a further 2.2 million cardiovascular deaths (WHO, 2017). Diseases of the exocrine pancreas includes acute pancreatitis, chronic pancreatitis, and pancreatic cancer. A recent meta-analysis reported that acute pancreatitis has the highest global incidence (33.7 cases per 100,000 person-years), while pancreatic cancer has the highest mortality rate (6.9 deaths per 100,000 person-years) (Xiao et al., 2016). Diseases of the exocrine pancreas lie on a continuum and there is strong evidence supporting a transition from acute to chronic pancreatitis via recurrence of the acute condition, driven by risk factors such as tobacco smoking and heavy alcohol consumption (Sankaran et al., 2015). Chronic pancreatitis has both endocrine and exocrine sequelae—diabetes of the exocrine pancreas and pancreatic exocrine insufficiency—in addition to causing a significantly increased risk of pancreatic cancer (Lew et al., 2017).

23217Alterations of pancreas size in diseases of the endocrine and exocrine pancreas have been discussed extensively in the literature, but few researchers have quantified the changes. Pancreas size can be quantified by noninvasive imaging via straightforward yet underused measurements such as pancreas volumes, diameters, areas, and lengths. While total pancreas volume provides a quantitative holistic measurement of pancreas size (DeSouza et al., 2018), the latter three measurements differ from it in several regards. First, they are typically applied to different regions of the pancreas, allowing regional alterations in size to be investigated. Various anatomical landmarks have been reported either to demarcate the different regions and/or to act as reference points for measurements. Second, the unidimensional parameters of pancreas diameter and length are measured along different axes, allowing for investigation into axial differences in pancreas size changes. Third, the latter three measurements are less time-consuming to obtain than total pancreas volume. Therefore, while these parameters do not provide a holistic assessment of pancreas size, they complement volumetry by providing information about regional alterations of pancreas size with regard to different axes. However, the clinical relevance of these parameters of pancreas size across a range of diseases has not been systematically reviewed to date.

The aims of the present study were to systematically document all studies investigating pancreas size using pancreas diameters, areas, and/or lengths in both diseased and normal human populations, to document the methodologies of these studies, in particular the anatomical landmarks used, and to amalgamate quantitative data about pancreas size across a range of diseases in a manner easily understandable by clinicians.

MATERIALS AND METHODS

RESULTS

Study Characteristics

The total number of articles retrieved from the three databases was 3,864. After duplicates had been removed (n = 1,574), 2,290 articles were screened on the basis of title and abstract. Full texts were obtained for 118 articles that were further screened for inclusion (Fig. 1). Finally, 28 studies (encompassing 3,810 individuals) met the eligibility criteria and were included in the systematic review (Table 1). Of these, 23 (encompassing 3,492 individuals) used transabdominal ultrasound (US), four (encompassing 215 individuals) used computed tomography (CT), and one (encompassing 103 individuals) used magnetic resonance (MR) imaging (Table 2).

Table 1. Characteristics of Included Studies

Study	Country	Type	Study groups	Total number of individuals (n)	Males n (%)	Age (mean ± SD), years	BMI (mean ± SD), kgm−2
Altobelli et al., 1998	Italy	Obs + CS	Normala	60	30 (50.0)	NRb	NR
			T1DMa	60	30 (50.0)	NRb	NR
Alzaid et al.,1993	Saudi Arabia	Obs + CS	Normal	19	4 (21.1)	37.0 ± 15.0	NR
			T1DM	14	11 (78.6)	26 ± 10	NR
			T2DM	43	22 (51.2)	51 ± 10	NR
Arger et al., 1979	USA	Obs + RC	Normal, AP, PCa, RPc	407d	NR	NR	NR
Basiratnia et al., 2007	Iran	Obs + CS	Normal	60	NR	NR	NR
			T1DM + T2DM	60	NR	NR	NR
Bilgin et al., 2009	Germany	Obs + RC	Normal	21	9 (42.9)	26.1 ± 4.3	NR
			CPe	82	54 (65.9)	60.2 ± 12.9	NR
Brunova et al.,1986	Czech Republic	Obs + CS	Normal	14	NR	39.1 ± 6.9	NR
			T1DMf	21	13 (61.9)	35.9 ± 8.1	NR
Coleman et al.,1983	USA	Obs + CS	Normala	70	NR	NRb	NR
			AP	25	NR	NRb	NR
Cuntz et al.,2000	Germany	Int + PC	Other psychiatric disorders	24	NR	27.6 ± 4.6	22.6 ± 3.8
			PEDg, h	109	2 (1.8)	26.1 ± 12.5	15.9 ± 4.1
d'Annunzio et al., 1996	Italy	Obs + CS	Normal	12	NR	NRi	NR
			T1DM	15	11 (73.3)	6.7 – 18.5j	NR
El-Hodhod et al., 2005	Egypt	Int + PC	Normal	12	5 (41.7)	1.2 ± 0.6	NR
			PEMk	33	15 (45.5)	1.0 ± 0.7	NR
Fonseca et al., 1985	UK	Obs + CS	Normal	19	NR	NR	NR
			T1DM	22	NR	NR	NR
			T2DM	19	NR	NR	NR
			T2DM + insulin	19	NR	NR	NR
Fleischer et al., 1983	USA	Obs + CS	Normal	17	NR	1.7– 16j	NR
			Pancreatitisl	21	NR	0.6 – 17.0j	NR
Gilbeau et al., 1992	Belgium	Obs + CS	Normal	57	27 (47.4)	60.5 ± 20 (SEM)	NR
			T1DM	20	10 (50.0)	48 ± 3.9 (SEM)	NR
			T2DM	25	9 (36.0)	61.7 ± 2.2 (SEM)	NR
			T2DM + Insulin	12	3 (25.0)	63.5 ± 9.9 (SEM)	NR
Migdalis et al., 1991	Greece	Obs + CS	Normal	80	42 (52.5)	51.1 ± 7.3	26.5 ± 3.2
			T2DM	84	44 (52.3)	52.4 ± 7.2	26.8 ± 3.6
Mohan et al., 1985	India	Obs + CS	Normal	20	12 (60.0)	23 ± 1.2	21.3 ± 1.0
			DEP	20	12 (60.0)	25 ± 2.3	19.6 ± 0.9
			MODY	20	11 (55.0)	26 ± 1.5	22.5 ± 1.5
			T1DM	20	11 (55.0)	24 ± 2.5	20.3 ± 1.2
Ohno et al., 1981	Japan	Obs + CS	Normal	10	NR	NR	NR
			T1DM	13	NR	NR	NR
			T2DM	13	NR	NR	NR
Oliviera et al., 2008	Brazil	Obs + CS	Normal	14	NR	NR	NR
			IGHD—GHRH-R mutant homozygotes	11	NR	NR	NR
Raeder et al., 2008	Norway	Obs + CS	Normal	11	7 (63.6)	9.4 ± 2.3	18.6 ± 2.9
			Nondiabetic MODY-8 CEL mutant homozygotes	11	7 (63.6)	9.5 ± 3.8	16.8 ± 2.5
Rajput et al., 2001	India	Obs + CS	Normal	15	NR	NR	NR
			T1DMm	35	NR	NR	NR
Siegel et al., 1987	USA	Obs + CS	Normala	273	NR	<0.1 – 19j	NR
			AP	13	NR	NRb	NR
Shawker et al., 1984	USA	Obs + CS	Normal	16	8 (50.0)	26.3 ± 10	NR
			CF	16	6 (37.5)	16.3 ± 5	NR
Silva et al., 1993	Brazil	Obs + CS	Normal	60	20 (33.3)	42.2 ± 16.7	22.7 ± 3.0
			T1DMm	40	20 (40)	26.5 ± 11.2	20.9 ± 3.0
			T2DMm	36	6 (16.7)	60.0 ± 12.6	26.4 ± 3.6
Soyer et al., 1999	France	Obs + CS	Normal	15	9 (60.0)	27 ± 8.3	NR
			CF	15	10 (66.7)	25 ± 6.5	NR
Tamura et al., 1988	Japan	Obs + CS	Normal	85	60 (70.6)	35.3 ± 9.2	NR
			T1DM	31	12 (38.7)	23.9 ± 7.2	NR
			T2DM-J	15	4 (26.7)	25.7 ± 8.4	NR
			T2DM-M	23	19 (82.6)	49.7 ± 11.4	NR
Taudorf et al., 1984	Denmark	Int + PC	Normal	10	NR	NR	NR
			Atopy	7	NR	NR	NR
Tosun et al., 2014	Turkey	Obs + CS	Normal pregnant women	230	0 (0)	28.1 ± 5.7	24.6 ± 2.9
			GDM	28	0 (0)	26.6 ± 5.3	24.7 ± 2.9
Treiber et al., 2016	Germany	Obs + CS	Normal	921	443 (48.1)	40.5 ± 12.6	23.7 ± 4.3
			CP	72	41 (56.9)	NR	NR
Valean et al., 1997	Romania	Obs + CS	Normal	108	NR	NRb	NR
			T1DM	27	NR	0.7 – 17j	NR

• Abbreviations: AP, acute pancreatitis; CEL, carboxyl ester lipase; CF, cystic fibrosis; CP, chronic pancreatitis; CS, cross-sectional; DEP, diabetes of the exocrine pancreas; GDM, gestational diabetes mellitus; GHRH-R, growth hormone releasing hormone receptor; IGHD, isolated growth hormone deficiency; Int, interventional; MODY, mature onset diabetes of the young; NR, not reported; Obs, observational; PC, prospective cohort; PCa, pancreatic cancer; RC, retrospective cohort; RP, retroperitoneal pathology; SEM, standard error of the mean; SD, standard deviation; T1DM, type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus; T2DM-J, juvenile onset T2DM; T2DM-M, mature onset T2DM.
• a Further stratified by age.
• b Age was not reported as a mean or by range, but this group comprised infants, children, and adolescents.
• c Comprises patients with primary retroperitoneal masses or metastatic retroperitoneal lymph nodes not of pancreatic origin, within the peri-pancreatic area.
• d This is the total number of individuals in the study; however, the total number of individuals per group is not reported, but rather the total number of ultrasound scans. Normal: 168 scans, AP: 236 scans, PCa: 35 scans, and RP: 16 scans.
• e Further stratified by the presence of diabetes mellitus and pancreatic exocrine insufficiency.
• f Further stratified by serum C-peptide serum concentrations.
• g Subgroups statistically combined.
• h Includes subgroups of anorexia nervosa (restrictive type), anorexia nervosa (bulimic type), bulimia nervosa (restrictive type), and bulimia nervosa (purging type).
• i Age was not reported as a mean or by range, but this group comprised children and adolescents.
• j Range.
• k Includes subgroups of kwashiorkor, marasmus and marasmic kwashiorkor malnutrition.
• l Comprises 19 patients with acute pancreatitis and two with chronic pancreatitis. The two individuals with chronic pancreatitis were not included in any analyses.
• m Further stratified by disease duration.

All 28 studies reported on pancreas size in diseased cohorts and respective control cohorts. Nineteen of them included a single cohort of diseased individuals and nine included multiple cohorts of different diseases (Table 1). Overall, nine disease settings were covered: diabetes mellitus [T1DM (12 cohorts), T2DM (seven cohorts), mature onset diabetes of the young (two cohorts), DEP (one cohort), gestational diabetes mellitus (one cohort), and mixture of T1DM and T2DM (one cohort)], pancreatitis [acute (four cohorts) and chronic alone (two cohorts)], cystic fibrosis (two cohorts), pancreatic cancer (one cohort), allergy to soy protein (one cohort), isolated growth hormone deficiency (one cohort), retroperitoneal pathologies (one cohort), protein energy malnutrition (one cohort), and psychiatric eating disorders (one cohort) (Table 1).

Pancreas Diameters

Diabetes mellitus

Ten studies investigated pancreas diameters in the setting of diabetes mellitus. Three of them included separate T1DM and T2DM cohorts (Ohno et al., 1981; Gilbeau et al., 1992; Silva et al., 1993), three included only T1DM cohorts (Brunova et al., 1986; Valean et al., 1997; Altobelli et al., 1998), one a gestational diabetes mellitus (GDM) cohort (Tosun et al., 2015), one a mature onset diabetes of the young type 8 (MODY-8) cohort (Raeder et al., 2008), one separate cohorts of diabetes of the exocrine pancreas (DEP), MODY (type unspecified), and T1DM (Mohan et al., 1985), and one a mixed T1DM and T2DM cohort (Basiratnia et al., 2007) (Table 3).

Table 3. Significance of Pancreas Size Change in Diseased Individuals Relative to Controls

• Abbreviations: ASP, allergy to soy protein; CF, cystic fibrosis; CP, chronic pancreatitis; DEP, diabetes of the exocrine pancreas; F, female; GDM, gestational diabetes mellitus; I, insulin; IGHD, isolated growth hormone deficiency; Juv, juvenile; M, male; Mat, mature; MODY, mature onset diabetes of the young; P, pancreatitis; PCa, pancreatic cancer; PED, psychiatric eating disorders; PEM, protein-energy malnutrition; RP, retroperitoneal pathology.
• a Differences between T1DM and T2DM individuals and controls were reported separately; however, the subject characteristics were reported for the mixed group of T1DM and T2DM.
• b No P values reported; however, isolated pancreas head and tail enlargement was more frequent than in controls.
• c Total pancreas size as measured by the combination of the AP diameters of the head, body, and tail, was significantly smaller in all groups of CP patients than controls (P < 0.0001).
• d Pancreas diameters, though they were measured, were not compared with controls.
• e Comprises a variety of nonpancreatic retroperitoneal masses that abutted the pancreas thereby increasing pancreas size measurement.
• f Head diameters used to calculate head volume. [Color table can be viewed at wileyonlinelibrary.com]

The most common finding from these studies was that pancreas diameters in individuals with T1DM were significantly lower than in controls (seven out of eight studies). Five studies measured diameters across the head, body, and tail; three of these measured AP diameters (Mohan et al., 1985; Brunova et al., 1986; Silva et al., 1993), one measured both AP and CC diameters (Altobelli et al., 1998), and the other did not specify the axis of measurement (Gilbeau et al., 1992). Among these five studies, three reported on reductions across all three regions (Gilbeau et al., 1992; Silva et al., 1993; Altobelli et al., 1998), one across just the body and tail (Brunova et al., 1986), and one reported no difference from controls (Mohan et al., 1985). Two studies measured AP diameters across the head and body only, and both reported significant reductions in these regions (Valean et al., 1997; Basiratnia et al., 2007). One study measured AP diameters across the body only and reported a significant reduction in this region (Ohno et al., 1981) (Table 3).

Four of the eight studies investigating pancreas diameters in the setting of T1DM also investigated them in the setting of T2DM. The most common finding they reported (three out of four studies) was that pancreas size in individuals with T2DM was intermediate between those with T1DM and controls. Two of these studies reported intermediate AP diameters across the body (Ohno et al., 1981; Gilbeau et al., 1992) and one across both the head and body (Basiratnia et al., 2007). However, one study reported no significant difference in pancreas diameters between T2DM individuals and controls overall (Silva et al., 1993).

The relationship between pancreatic function, both endocrine and exocrine, and pancreas diameter in the setting of T1DM and T2DM was investigated in four studies (Ohno et al., 1981; Brunova et al., 1986; Gilbeau et al., 1992; Altobelli et al., 1998). Pancreatic endocrine function was assessed via glycemic control (by proxy of glycated hemoglobin and the nature of therapy) in two studies (Gilbeau et al., 1992; Altobelli et al., 1998), and by insulin secretion (by proxy of C-peptide) in all four. Glycemic control by glycated hemoglobin did not correlate significantly with pancreas diameter in individuals with T1DM (Altobelli et al., 1998); however, individuals with T2DM requiring exogenous insulin for glycemic control had lower AP diameters across all regions of the pancreas than T2DM individuals treated by diet and oral hypoglycemics (Gilbeau et al., 1992). This same study was also one of the three that reported on insulin secretion and undertook subgroup analyses suggesting a positive relationship between C-peptide and pancreas diameters across both T1DM and T2DM (Ohno et al., 1981; Brunova et al., 1986; Gilbeau et al., 1992). In contrast to the findings of these three studies, one study (undertaking correlation analyses) reported that the relationship between C-peptide and pancreas diameter was not significant in their T1DM cohort. Pancreas exocrine function was only investigated in one study via a pancreatic function test; pancreatic amylase, trypsin, and chymotrypsin were found to be lowest in individuals with the greatest reduction in AP diameter of the body (Ohno et al., 1981).

The relationship between disease duration and pancreas diameter in T1DM and T2DM was investigated in five studies (Brunova et al., 1986; Gilbeau et al., 1992; Silva et al., 1993; Altobelli et al., 1998; Basiratnia et al., 2007). Two of these undertook correlation analyses and both reported significant negative correlations between disease duration and pancreas size across both types of diabetics, one reporting the correlation with head AP diameters (Basiratnia et al., 2007) and the other with tail AP diameters (Gilbeau et al., 1992). The remaining three studies conducted subgroup analyses based on various disease durations. Adolescents between the ages of 13 and 15 years with a T1DM duration greater than five years had the greatest reduction in AP and CC diameters across all regions, and had pancreata that were smaller than those of children between the ages of two and three years who had a T1DM duration less than two years (Altobelli et al., 1998). This finding was reflected in adults with T1DM: individuals with disease duration greater than ten years showed the greatest reduction in AP diameter across all regions of the pancreas in one study (Silva et al., 1993) and across the body and tail in another (Brunova et al., 1986). In contrast, the only study to stratify a T2DM cohort by disease duration reported that pancreas size differed significantly from controls only in a greater body AP diameter after ten years (Silva et al., 1993).

The other three diseases in the diabetes mellitus spectrum that were covered were DEP (Mohan et al., 1985), GDM (Tosun et al., 2015), and MODY (Mohan et al., 1985; Raeder et al., 2008). The results of these studies are presented in Table 3.

Pancreatitis

Six studies investigated pancreas diameters in the setting of pancreatitis: four included patients with acute pancreatitis (Arger et al., 1979; Coleman et al., 1983; Fleischer et al., 1983; Siegel et al., 1987) and two included cohorts of chronic pancreatitis only (Bilgin et al., 2009; Treiber et al., 2016) (Table 3).

Three of the four studies that investigated pancreas diameters in acute pancreatitis reported that enlargement was the most common finding (Arger et al., 1979; Coleman et al., 1983; Fleischer et al., 1983). Two of these studies comprised small cohorts and were based on pediatric populations (Coleman et al., 1983; Fleischer et al., 1983). One measured body AP diameter (adjusted by lumbar vertebral body transverse diameter) and reported that it was significantly larger in children with acute pancreatitis than controls (Fleischer et al., 1983). The other study reported that 60% of children with pancreatitis displayed focal enlargement of the pancreas (86.7% of whom displayed enlargement of the head only and 13.3% of the tail) (Coleman et al., 1983). Notably, the study that reported no significant difference in overall pancreas size between children with acute pancreatitis and controls also reported that focal head enlargement was most commonly seen in those with enlargement of the gland (Siegel et al., 1987). The only study to investigate acute pancreatitis in an adult population reported that head, body, and tail AP diameters were significantly larger in acute pancreatitis patients than controls. Further, head, body, and tail AP diameters exceeded 2.5 cm, 2.0 cm, and 2.0 cm, respectively, in a significantly greater proportion of individuals with pancreatitis than controls [acute pancreatitis (59%, 67%, and 71%) versus controls (0%, 6%, and 3%)] (Arger et al., 1979).

The two studies that investigated pancreas diameters in the setting of chronic pancreatitis alone gave conflicting results. One reported that AP diameters across all three regions of the pancreas were lower in individuals with chronic pancreatitis than controls (Bilgin et al., 2009), whereas the other reported that the AP diameters of the head and body, and the diameters perpendicular to the midline of the tail, were significantly greater in individuals with chronic pancreatitis than controls (Treiber et al., 2016). The latter study also specified that this increase in pancreas diameter was still significant in individuals who had not experienced an acute bout of pancreatitis and had disease duration greater than two years (Treiber et al., 2016).

One study assessed pancreatic exocrine function in the setting of chronic pancreatitis (Bilgin et al., 2009) through fecal elastase-1 tests and reported a significant positive correlation between pancreas size (mean of head, body, and tail AP diameters) and fecal elastase-1. Furthermore, this cohort was stratified into three subgroups: isolated diabetes mellitus (DM), isolated pancreas exocrine insufficiency (PEI), and combined diabetes mellitus and pancreas exocrine insufficiency (DM + PEI). Individuals with DM + PEI had a significantly smaller pancreas size than those with DM alone. Individuals with DM alone did not differ significantly in pancreas size from those with PEI alone. Although it was specified that DM was diagnosed through retrospective review, it was not specified whether those diagnoses were made prior to or after the onset of chronic pancreatitis symptoms.

Pancreatic cancer

One study investigated pancreas diameters in the setting of pancreatic cancer (Arger et al., 1979). It included cohorts of individuals with pancreatic cancer, acute and chronic pancreatitis, retroperitoneal pathologies, and controls (Table 3). The retroperitoneal pathologies comprised primary retroperitoneal masses or metastatic retroperitoneal lymph nodes not of pancreatic origin in the peri-pancreatic area.

Individuals with pancreatic cancer showed both focal and diffuse enlargement of the pancreas. Head, body, and tail AP diameters exceeded 2.5 cm, 2.0 cm, and 2.0 cm in a significantly greater proportion of individuals with pancreatic cancer than controls [pancreatic cancer (80%, 74%, and 42%) versus controls (0%, 6%, and 3%)] (Arger et al., 1979). Further, individuals with pancreatic cancer had a greater frequency of head and body enlargement than those with pancreatitis and retroperitoneal pathologies. Tail enlargement was more frequent in individuals with retroperitoneal pathologies (50%), but this was attributable to poor delineation of the normal tail from surrounding pathology.

However, this study did not report the type of pancreatic cancer (e.g., pancreatic ductal adenocarcinoma versus neuroendocrine tumors etc.) or its staging. Further, the region in which the primary mass was located was not reported, nor were the frequencies of individuals presenting with isolated focal enlargement (due to the primary mass) and/or diffuse enlargement. Finally, it did not specify whether the mass was excluded from pancreas diameter measurements.

Other diseases

Pancreas diameters were also investigated in the settings of cystic fibrosis (Shawker et al., 1984; Soyer et al., 1999), protein-energy malnutrition (El-Hodhod et al., 2005), psychiatric eating disorders (Cuntz et al., 2000), and soy allergy (Taudorf et al., 1984). The results of these studies are presented in Table 3.

DISCUSSION

This study has systematically reviewed the current literature on pancreas size as assessed by pancreas diameters, areas, and lengths. Its strengths include a robust search strategy, the systematic documentation of anatomical landmarks, and the amalgamation of quantitative data regarding alterations in pancreas size across a range of diseases.

The anatomical landmarks used for pancreas size measurements have been systematically documented. These landmarks were used either to demarcate the pancreas into three regions or to act as reference points for diameter measurements. There was consistency among studies that used anatomical landmarks to demarcate regions of the pancreas, where the confluence of the superior mesenteric and splenic veins served to divide the head (to its right) and the body (to its left), and the tail was defined as the region immediately anterior to the left kidney. In contrast, there was significant heterogeneity among studies using landmarks as reference points for diameter measurements. The confluence of the superior mesenteric and splenic veins, the superior mesenteric artery, and the medial border of the left kidney served most commonly as the landmarks of reference for diameter measurements across the head, body, and tail, respectively. The potential benefit of using anatomical landmarks is that they provide an easy stepping stone for standardizing future investigations into regional alterations in pancreas size. However, this benefit will only be achieved when two major drawbacks of the current field are overcome.

The first drawback is the significant variability associated with measurements of pancreas size, which could be influenced by patient, technical, and rater-related factors. To measure pancreas size, the pancreas and its anatomical landmarks must be visualized in the same plane. Visualization can be impaired by patient factors including the presence of bowel gas, ascitic fluid, peri-pancreatic masses, and minimal visceral fat. Furthermore, the overall position of the pancreas and the direction of its regions (in particular, the tail) relative to the position of anatomical landmarks can differ among individuals. For example, individuals with a tail that moves in a postero-superior direction relative to the body are more likely to have fewer common axial slices with the left kidney than individuals with a tail that moves in a postero-inferior course relative to the body. Technical factors specific to US (transducer frequency, the use of water in the stomach as an acoustic window, or the application of Doppler imaging) and cross-sectional imaging in CT/MR (such as the plane of sectioning, the administration of contrast, and the pulse sequences used for MR) can also affect visualization. Rater-related factors such as knowledge and experience of abdominal anatomy in the context of the specific imaging method could also contribute to variability.

Compounding the drawback of significant variability are the issues of poor inter-rater and intra-rater reliability attributable to the nature of the measurements themselves. Unidimensional measurements such as pancreas diameter and length are likely to be more variable than pancreas area measurements, since the angle of the measurement can significantly affect the value obtained. This is important as pancreas diameters have been measured either strictly along the anteroposterior (AP) or craniocaudal (CC) axes, or in relation to the organ midline (either perpendicular or parallel to it). While perpendicular measurements have the benefit of accounting for the course of the organ, the freedom in the angle of the measurement can lead to poor inter-rater reliability. This contrasts with AP or CC diameters, where the angle of measurement is set at 90 degrees. The use of anatomical landmarks to define measurements is also significant as inter-rater reliability can differ between diameters based on reference anatomical landmarks and measurements of the greatest diameter in the region. The initiation points and axes deemed most appropriate by the raters to obtain the ‘greatest’ diameter could be more variable. The problem of selection of the initial point is further complicated in lobulated pancreata with variant courses. To date, only one study has compared inter-rater reliability between greatest diameters and diameters based on anatomical landmarks, but those measurements were limited to the head only (Cuntz et al., 2000). Future studies investigating inter-rater reliability in diameter measurements across all regions of the pancreas are required.

The second drawback is the lack of critical appraisal of published anatomical landmarks. The authors of primary studies did not justify their choice of particular anatomical landmarks rather than previously published landmarks. To date, no one has conducted a structured objective assessment of the different anatomical landmarks reported in the literature. The findings of such a study would provide a sound basis for guidelines recommending anatomical landmarks for future studies. In the interim, however, the following considerations could help in determining anatomical landmarks for diameter measurements (Table 4).

The inferior vena cava, the confluence of the superior mesenteric and splenic veins, the superior mesenteric vessels, and the hepatic portal vein have been used as references for diameter measurements of the head. Despite the difficulty in visualizing the venous confluence (as compared to the other landmarks), this landmark yields a diameter that is more centered within the head than the alternatives. Thus, a greater proportion of the head's mass is likely to be covered. The midline of the superior mesenteric artery and the left lateral border of the lumbar vertebrae have been used as references for measurements of the body diameter. Given that the pancreas body is superior to the origin of the superior mesenteric artery, it is unlikely that the two structures will be visible together in an axial view, so it might not be a reliable landmark for cross-sectional imaging. In contrast, the left lateral border of a lumbar vertebral body would be a more reliable landmark for the body given that lumbar vertebrae span the posterior wall of the abdomen. The internal border of the left kidney and the left lateral border of the lumbar vertebrae have been used as references for diameter measurements of the tail. A perpendicular diameter across the tail in line with the internal border of the left kidney would cross a region that would certainly be considered as tail. However, using the left lateral border of a lumbar vertebral body, one could argue that the region covered is body rather than tail. Therefore, a more reliable anatomical landmark for the tail would be the internal border of the left kidney.

This study has also amalgamated quantitative data regarding alterations in pancreas size across a range of diseases. Pancreas size differed significantly between diseased individuals and controls. Insulin-dependent individuals with diabetes mellitus (those with T1DM and those with T2DM requiring exogenous insulin therapy) had a significantly smaller pancreas than controls, the extent of reduction being greater in individuals with T1DM than those with T2DM requiring exogenous insulin. Further, there were regional differences in the extent of parenchymal atrophy with regard to disease duration in both T1DM and T2DM. Overall, it appears that the tail is the first to atrophy and the head is the last. These findings can be explained by the ‘insulino-acinar hypothesis,’ which proposes that the growth and repair of the gland's exocrine tissue depends on local concentrations of insulin (a potent trophic hormone). Therefore, a reduction in β-cell insulin secretion and/or altered local insulin receptor activity would lead to parenchymal atrophy (Foulis and Stewart, 1984; Gilbeau et al., 1992). This could also explain the commonly-reported positive correlation between C-peptide and pancreas size. Further, islet density is reportedly greater in the body and tail (Rahier et al., 1983; Wang et al., 2013). Therefore, insulitis could have more profound effects on these regions in T1DM, resulting in greater atrophy. Future autopsy studies employing modern immunohistochemistry and imaging are required to test this prediction, as previous smaller autopsy studies have given inconsistent results.

In T1DM, parenchymal atrophy can be explained by stark insulinopenia secondary to autoimmune-based insulitis, a phenomenon evidenced biochemically by auto-antibodies against β-cells and insulin (Taplin and Barker, 2008) and by postmortem findings of lymphocyte infiltration of the islets (Foulis and Stewart, 1984). The intermediate pancreas size in T2DM is difficult to explain given the heterogeneous etiology of this condition, which involves components of fatty infiltration (Singh et al., 2017a, 2017b) and peripheral insulin resistance. Further, its pathophysiology in terms of alterations in pancreas size could depend on the time course of the disease, as evidenced by a negative correlation between total pancreas volume and disease duration (DeSouza et al., 2018). While prospective studies are required to prove this, the earlier stages of T2DM could be characterized by increases in pancreas size due to a combination of fatty infiltration and hyperinsulinemia secondary to peripheral insulin resistance. However, as the disease progresses and glycemic control worsens, parenchymal atrophy could ensue as a result of growing insulinopenia from β-cell dysfunction and subsequent failure. Therefore, the differences in findings among studies could in part be attributed to cohorts heterogeneous in terms of disease duration.

In most studies investigating acute pancreatitis, pancreas size was significantly greater than in normal individuals. Pancreas enlargement, both focal and diffuse, was the most common finding in this condition. This could be explained by interstitial edema due to extravasation as part of the inflammatory response to autodigestion of the gland (Wang et al., 2009). Focal enlargement of the pancreas head was commonly reported. Perhaps this could be due to regional differences in the extent of pancreas enlargement among different disease etiologies. For example, individuals with biliary acute pancreatitis could have a higher rate of and/or a greater extent of head enlargement owing to more extensive inflammation of tissues immediately surrounding the mechanical obstruction of intrapancreatic biliary structures such as the distal common bile duct and the hepatopancreatic ampulla. However, this regional difference in pancreas enlargement could have a time course and could diminish after decompression through natural breakdown, endoscopic retrieval, or surgical removal of the causative gallstone. In contrast, alcohol-induced acute pancreatitis can be characterized by diffuse enlargement of the gland associated with widespread inflammation in response to the systemic toxin being carried throughout the gland via its extensive vascular network. However, none of the studies reported pancreas size in relation to the etiology of acute pancreatitis. Future studies that stratify their acute pancreatitis cohorts by etiology are warranted to investigate this hypothesis.

The two studies of pancreas size in chronic pancreatitis were inconsistent: one reported that it was reduced (Bilgin et al., 2009) and the other that it was increased (Treiber et al., 2016). This discrepancy could partly be explained by differences in the disease durations of the study populations; only one of these studies reported the mean disease duration of their chronic pancreatitis cohort (Treiber et al., 2016). This is relevant as pancreas size could change through the time course of the disease. The pancreas could be enlarged by mechanisms similar to those in acute pancreatitis during acute exacerbations of early chronic pancreatitis. However, as the disease progresses with atrophy and fibrous replacement of parenchymal tissue, the pancreas could begin to shrink. These findings would be consistent with the pathogenesis of chronic pancreatitis, which involves feed-forward activation of pancreatic stellate cells by TGF-β released by activated macrophages, leading to the deposition of collagen and extracellular matrix proteins to replace parenchyma destroyed by auto-digestion (Lew et al., 2017). The ‘insulino-acinar hypothesis’ could also explain parenchymal atrophy in conjunction with the processes described above, as β-cell dysfunction secondary to chronic inflammation and ectopic fat accumulation could lead to subclinical insulinopenia in early CP resulting in atrophy of the exocrine tissue (Schrader et al., 2009). The diagnosis of early chronic pancreatitis is challenging, and quantification of parenchymal atrophy during periods of senescence via pancreas diameters, areas, and lengths in early CP could offer a new avenue for early diagnosis since traditional radiological findings (such as parenchymal and intraductal calcification, dilatation of the main pancreatic duct and side branches, and contour lobulation) are typically found only in late disease (Sarner and Cotton, 1984).

The main limitation of this review is that these quantitative measures of pancreas size were not amenable to meta-analysis. This is because of the marked methodological heterogeneity in the primary studies. There is now a need to establish guidelines based on modern imaging studies to allow anatomical landmarks to be standardized for future studies. Once this has been achieved, meta-analyses can be conducted. There is also a need for more population-based studies to re-define norms for pancreas size as stratified by age, BMI, ethnicity, and sex on the basis of established guidelines using modern imaging. While several studies using US and CT have reported pancreas size exclusively in normal individuals (Weill et al., 1977; de Graaff et al., 1978; Heuck et al., 1987; Ueda, 1989; Capaccioli and Carpi, 1993; Giandomenico et al., 1993; Nyren et al., 1996; Chen et al., 1997; Djuric-Stefanovic et al., 2012), most of such studies did not use modern imaging methods, which have significantly improved resolution.

In conclusion, the importance of quantifying pancreas size via noninvasive imaging is being increasingly recognized in the context of early diagnosis and management of diseases of the endocrine and exocrine pancreas. This systematic review has documented the methodologies of studies investigating pancreas size using diameters, areas, and lengths, and in particular it has documented the various anatomical landmarks used. Further, quantitative data regarding alterations in pancreas size across several diseases have been amalgamated in a manner easily understandable by clinicians. Pancreas diameters and areas were reduced in individuals with diabetes mellitus (more markedly in T1DM than T2DM) and pancreas diameters were increased in individuals with diseases of the exocrine pancreas (with the exception of chronic pancreatitis). Future prospective studies based on modern imaging and standardized protocols are now warranted to re-define norms of pancreas size, and to elucidate the effects of differences in disease duration and etiology on the endocrine and exocrine pancreas.

DECLARATION OF CONFLICT OF INTEREST

The authors declare no conflicts of interest.