Endoscopic ultrasound in chronic pancreatitis: Where are we now?

Abstract

Endoscopic ultrasonography (EUS) is well suited for assessment of the pancreas due to its high resolution and the proximity of the transducer to the pancreas, avoiding air in the gut. Evaluation of chronic pancreatitis (CP) was an early target for EUS, initially just for diagnosis but later for therapeutic purposes. The diagnosis of CP is still accomplished using the standard scoring based on nine criteria, all considered to be of equal value. For diagnosis of any CP, at least three or four criteria must be fulfilled, but for diagnosis of severe CP at least six criteria are necessary. The Rosemont classification, more restrictive, aims to standardize the criteria and assigns different values to different features, but requires further validation. EUS-fine needle aspiration (EUS-FNA) is less advisable for diagnosis of diffuse CP due to its potential side effects. Elastography and contrast-enhanced EUS are orientation in differentiating a focal pancreatic mass in a parenchyma with features of CP, but they cannot replace EUS-FNA. The usefulness of EUS-guided celiac block for painful CP is still being debated with regard to the best technique and the indications. EUS-guided drainage of pseudocysts is preferred in non-bulging pseudocysts or in the presence of portal hypertension. EUS-guided drainage of the main pancreatic duct should be reserved for cases in which endoscopic retrograde cholangiopancreatography has failed owing to difficult cannulation of the papilla or difficult endotherapy. It should be performed only by highly skilled endoscopists, due to the high rate of complications.

Key Words: Endoscopic ultrasonography; Pancreatic neoplasms; Chronic pancreatitis; Contrast agents; Nerve block; Pancreatic pseudocyst; Drainage; Elastography; Main pancreatic duct

INTRODUCTION

Chronic pancreatitis (CP) is an irreversible and progressive inflammatory process featuring pathological modifications of fibrosis, inflammatory infiltration, and destruction of exocrine and endocrine tissue, resulting in characteristic morphological changes in the parenchyma and pancreatic ducts. These modifications vary in intensity and distribution (diffuse or patchy). This has several consequences: (1) Biopsy specimens are difficult to obtain and not always relevant, because they do not fully display the signs of CP; moreover, duct biopsy is usually avoided due to the risk of acute pancreatitis; (2) Most imaging methods reflects only partially the CP modifications, especially those typical for late stages of the disease; some methods, such as endoscopic retrograde cholangiopancreatography (ERCP) and magnetic retrograde cholangiopancreatography (MRCP) detect only the ductal features of CP; and (3) The findings of pancreatic function tests are not modified until a late stage in the natural history of the disease. Endoscopic ultrasonography (EUS) accomplishes the quality of being an imaging method able to detect both early and late changes in the parenchyma and pancreatic ducts.

The pancreas is well assessed by EUS due to the method’s high resolution and the proximity of the transducer to the pancreas with the possibility of avoiding air in the gut. In patients with CP, EUS was performed initially for diagnosis, then for differential diagnosis, and later for therapeutic purposes (Figure 1).

Figure 1 Flowchart of the endoscopic ultrasonography utility in chronic pancreatitis. EUS: Endoscopic ultrasonography; CP: Chronic pancreatitis; CE-EUS: Contrast enhanced endoscopic ultrasonography; MPD: Main pancreatic duct.

POSITIVE DIAGNOSIS

Despite its advantage of assessing the pancreas at very close range, EUS, being operator dependent, is still imperfect in establishing the diagnosis of chronic pancreatitis. The various pathological aspects of the disease are shown as different EUS features, and the same importance for diagnosis has been attributed to all of them. There have been several attempts to define the disease on ductal and parenchymal criteria, initially embracing 11 criteria[1,2], then focusing on nine factors corresponding to histopathological changes[3]: five parenchymal criteria (hyperechoic foci, hyperechoic strands, parenchymal lobularity, cysts, calcifications) and four ductal criteria (pancreatic duct dilatation, pancreatic duct irregularity, hyperechoic pancreatic duct walls, visible pancreatic side branches) (Figure 2). Very rarely are all these manifestations present simultaneously. Some of these features have been found also in elderly people[4], males (OR = 1.8, 95% CI: 1.3-2.55), persons with a history of alcohol consumption abuse (OR = 5.1, 95% CI: 3.1-8.5), smokers (OR = 1.7, 95% CI: 1.2-2.4), and those with history of acute pancreatitis[5-9]. Some features, like gland atrophy or lobularity aspect, can impede the complete assessment of all features (e.g. visualization of side branches of pancreatic ducts).

Figure 2 Chronic pancreatitis. Parenchymal and ductal pancreatic stones as hyperechoic structures with shadowing and stenosis of the main pancreatic duct.

The interobserver agreement in one study using these criteria was moderate (k = 0.45), with good agreement only for duct dilatation and lobularity; the main drawback of the study was the limited experience of some examiners with pancreatic EUS. The most important criterion for the diagnosis was considered by all experts to be pancreatic stones, followed by visible side branches and lobularity, and the least significant was main pancreatic duct (MPD) dilatation[9]. In an EUS study in which both digital linear and radial echo endoscopes were employed, the interobserver variability also moderate (k = 0.50 and 0.61 respectively); the best concordance between the two methods was found for detection of cysts, calcifications, and visible side branches[10].

Because histological evaluation of the pancreas is usually difficult, different gold standards have been used to establish the optimum number of EUS criteria for diagnosis of CP. The secretin direct pancreatic test has 85% sensitivity and 85% specificity for CP diagnosis, and the false-negative results are due to preserved pancreatic exocrine function[11]. Using one or two criteria for mild pancreatitis, three to five for moderate pancreatitis, and more than five for severe forms, the agreement with the secretin test as gold standard was 100% for normal parenchyma and severe disease, 50% for moderate forms, and 13% for mild disease[2]. On comparison of both EUS radial and linear assessment with the endoscopic secretin test during the same procedure, the best EUS accuracy was obtained for a cut-off point of more than four criteria (accuracy of 84% and 74%, respectively)[10]. The same group obtained lower sensitivity and specificity for diagnosis using four EUS criteria when cholecystokinin was used instead of secretin to test pancreatic function[12]. Comparison of assessment by non-blinded EUS (three to five criteria for diagnosis) and endoscopic retrograde cholangiopancreatography (ERCP; Cambridge classification) showed quite similar sensitivity (72% vs 68%) and specificity (76% vs 79%) for either mild or severe chronic pancreatitis, with the secretin endoscopic direct pancreatic test as the reference. However, the odds ratio for exocrine insufficiency was higher for EUS assessment than for ERCP[13]. To obtain the best specificity and the best negative predictive value for diagnosis, six criteria were needed, however, the sensitivity was only 26%[8,14]. Secretin-stimulated EUS detected the features of CP better than EUS without secretin (12/13 patients) and the sensitized EUS seemed to be able to predict a favorable outcome or success of endoscopic treatment[15] (Table 1).

Table 1 Diagnostic value of endoscopic ultrasonography in chronic pancreatitis.

Author	No. of pts	No. of EUS criteria	Threshold for CP diagnosis	Comparison	Sn	Sp	PPV	NPV	Acc
Wiersema et al[4]	69	11	> 3 = dg	EUS vs ERCP	100	79
				EUS vs ERCP + secretin test	70	33
				EUS vs ERCP + history	90	66
Catalano et al[2]	80	11	1-2 mild	EUS vs secretin test	84	78
			3-5 moderate	EUS vs ERCP	86.1	95.4
			> 5 severe	EUS vs ERCP + secretin test	84.2	97.6
Sahai et al[8]	126	9	> 2 for any CP	EUS vs ERCP			> 85	< 85
			< 3 = fibrosis
			> 6 = severe				> 85
Conwell et al[14]	56	9	4-5 = equivocal	EUS vs ePFT	36	94	93	41
			> 6 = definite		26	100	100	39
Stevens et al[13]	83	9	3-5 = dg	EUS vs ERCP	68	79	83	62
			6-9 = severe
Stevens et al[10]	100	9	> 4	Radial EUS vs ePFT	68	95			84
				Linear EUS vs ePFT	44	95			74
Stevens et al[12]	50	9	> 4	EUS vs secretin ePFT	71	92
				EUS vs CCK ePFT	63	85
Zimmermann et al[23]	21	9	> 4	EUS vs histology (surgery)	78	73
Varadarajulu et al[24]	21	9	> 4	EUS vs histology1 (surgery)	90	85.7			88.1
Chong et al[25]	71	9	> 3 = dg	EUS vs histology1 (surgery)	83.3	80
			> 4 = severe fibrosis
Bhutani et al[22]	11	9	> 3	EUS vs histology (autopsy)

¹Non-calcific chronic pancreatitis. ePFT: Endoscopic pancreatic function test; EUS: Endoscopic ultrasonography; ERCP: Endoscopic retrograde cholangiopancreatography; Sn: Sensitivity; Sp: Specificity; Acc: Accuracy; CCK: Cholecystokinin; PPV: Positive predictive value; NPV: Negative predictive value.

Using ERCP as gold standard, more than two criteria or three criteria, respectively, were found to be optimal for diagnosis[4,8]. The EUS sensitivity for diagnosis varied between 68% and 100% and the specificity was 78%-97% when ERCP was considered the gold standard (Table 1). The overall agreement with ERCP was k = 0.51, but the concordance for mild forms on EUS was only 83%. The factors most predictive for abnormal ERCP were ductal stones and parenchymal calcifications[4]. Among patients with a normal pancreatogram, 84.2% were found to have parenchymal changes of CP (accentuation of lobular pattern, focal areas of reduced echogenicity, hyperechoic foci) or increased ductal wall echogenicity. During follow-up (median 18 mo), 68% of patients with initially normal findings on ERCP progressed to an abnormal pancreatogram, supporting the importance of EUS description for early CP. However, this evolution was not confirmed in a second study of alcoholic chronic cirrhosis and CP[16,17]. Evaluation of images can be improved by computer-assisted image analysis[18].

The patient’s history may be suggestive of CP. More than five features of CP were seen in 49.9% of 156 patients with persistent or non-specific dyspepsia[19]. Another study showed that there were more criteria for CP in the group with pain and steatorrhea than in the group with pain but no steatorrhea, so they concluded that history can be helpful in diagnosing CP[20].

Pathologic diagnosis, the ideal gold standard, is rarely obtained from surgical specimens, EUS fine needle aspiration (EUS-FNA) or Tru-Cut core biopsies. The correspondence of EUS criteria to pathologic changes is shown in Table 2[21,22]. One recent paper showed that in postmortem pancreatic specimens the presence of more than three EUS standard criteria of CP correlated with the histologic diagnosis, but these features were also present in elderly persons dying of diseases other than CP[22] and in 59% of asymptomatic alcohol abusers[5].

Table 2 Correspondence between standard endoscopic ultrasonography criteria and pathologic features in chronic pancreatitis (adapted from Sahai AV 2002[21]).

Standard EUS criteria	Pathologic features
Parenchymal criteria
Hyperechoic foci	Small calcifications
Hyperechoic strands	Fibrosis
Lobularity	Edema or fibrosis
Cysts	Pseudocysts
Calcifications	Calcifications
Ductal criteria
MPD dilatation	MPD dilatation
MPD irregularity	MPD irregular
Hyperechoic MPD walls	Ductal fibrosis or edema
Visible side branches	Dilated secondary branches

EUS: Endoscopic ultrasonography; MPD: Main pancreatic duct.

Comparing the EUS standard criteria with the histologic findings from specimens obtained during surgery, fulfillment of five or more criteria was associated with sensitivity of 60% and specificity of 83%, compared with 87% and 64% respectively when three criteria were used[23]. Good correlation with histology obtained during surgery of non-calcific CP was also found for the presence of four pancreatic features and for EUS findings of foci, stranding, lobulation, or ductal modifications. A limitation of this study was its use of surgical specimens secondary to neoplastic pancreatic disease[24]. Using surgical specimens obtained after preoperative EUS, three criteria were shown to differentiate abnormal from normal pancreatic tissue, but four criteria represented the limit for identification of severe fibrosis[25]. Again, the use of four EUS criteria compared with the association of ERCP, surgical pathology, and/or long-term clinical follow-up showed that EUS was more sensitive than MRCP but equally specific, and when both tests were abnormal the specificity was 100%[26]. Therefore, three or four criteria seems to suffice to rule out CP, but to establish the diagnosis at least six criteria are necessary[27].

The diagnosis of autoimmune pancreatitis is based on the same criteria, but for early stages (corresponding to Cambridge grade 0 to 2) the characteristic criteria are lobularity and hyperechoic pancreatic duct walls[28]. One study found diffuse hypoechoic areas, diffuse enlargement of the parenchyma, focal hypoechoic areas, and bile duct wall thickening as supplementary features characterizing autoimmune pancreatitis; these manifestations resolved after steroid treatment and were helpful in differentiation from ductal adenocarcinomas[29]. EUS-FNA is able to show a stromal structure with high lymphoid cellularity[30]. Lymphoplasmocytic sclerosing pancreatitis can be more accurately detected in tissue samples obtained by Tru-Cut biopsy[31]. With regard to the assessment of severity, preliminary data have pointed to significant diagnostic EUS features: hyperechoic foci for mild CP; hyperechoic foci, visible side branches, and duct dilatation for moderate CP; and visible side branches, duct dilatation, duct irregularity, and calcifications for severe CP[32].

Because the different pathological characteristics of CP vary in importance, the nine-criteria scheme assigning each criterion the same importance is insufficiently reliable and its diagnostic accuracy doubtful. The Rosemont classification, elaborated by international consensus, uses parenchymal and ductal criteria divided into major and minor features (Table 3). On this basis the findings are classified as “consistent with CP”, “suggestive of CP”, “indeterminate for CP”, or “normal” (Table 4)[33]. This system, quite complicated and more restrictive in diagnosing CP, proved to agree with the diagnostic classification of the nine-criteria scheme in 74% of cases, increasing to 84% when “suggestive of CP” was included[34,35]. Using this system, the findings were similar for radial and linear EUS, with good results for parenchymal criteria (cysts 100%, hyperechoic foci 98%, lobularity/dilated ducts 94%) and modest results for dilated side branch, irregular pancreatic duct and hyperechoic wall of MPD[36]. In a recent multicenter study, 14 experts evaluated 50 recorded videos using the standard nine EUS criteria (diagnostic: > 4 criteria) and the Rosemont criteria (diagnostic: suggestive of CP or consistent with CP). They obtained substantial interobserver agreement for the Rosemont classification (k = 0.65) and moderate agreement for the standard classification (k = 0.54); the difference was not significant. The best agreement was noted for calcifications (standard scoring), pancreatic duct calcifications (Rosemont classification) and pancreatic duct dilation (both systems). The least agreement was seen for lobularity without honeycomb (Rosemont classification). This study used computed tomography (CT) and endoscopic pancreatic function test (ePFT) as gold standard, without histology. The patients were correctly classified as “definite CP” in 91.2% of cases (standard scoring) and 83.5% (Rosemont scoring); as “mild CP” in 50% (standard scoring) and 42.9% (Rosemont scoring); and “no CP” in 83.3% and 95.2% of cases respectively[37]. Further validation of the Rosemont classifications is needed.

Table 3 Rosemont consensus definitions.

Rank	Features		Definition	Location
	Parenchymal features
1	Major A	Hyperechoic foci with shadowing	Echogenic structures ≥ 2 mm in length and width that shadow	Body and tail only
2	Major B	Lobularity with honeycombing	Well-circumscribed, ≥ 5 mm structures with enhancing rims and relatively echo-poor centers, with ≥ 3 lobules	Body and tail only
	Minor	Lobularity with honeycombing	Well-circumscribed, ≥ 5 mm structures with enhancing rims and relatively echo-poor centers, with non-contiguous lobules	Body and tail only
3	Minor	Hyperechoic foci without shadowing	Echogenic structures ≥ 2 mm in length and width with no shadowing	Body and tail only
4	Minor	Cysts	Anechoic, rounded/elliptical structures with or without septations	Head, body and tail only
5	Minor	Stranding	Hyperechoic lines ≥ 3 mm in length in at least two different directions with respect to the imaged plane	Body and tail only
	Ductal features
1	Major A	MPD calculi	Echogenic structures within the MPD with acoustic shadowing	Head, body and tail only
2	Minor	Irregularity of MPD contour	Uneven or irregular outline and ectatic course	Body and tail only
3	Minor	Dilated side branches	3 or more tubular anechoic structures each measuring ≥ 1 mm in width, budding from MPD	Body and tail only
4	Minor	MPD dilation	≥ 3.5 mm in body or > 1.5 mm in tail	Body and tail only
5	Minor	Hyperechoic duct margin	Echogenic, distinct structure greater than 50%of the entire MPD	Body and tail only

MPD: Main pancreatic duct.

Table 4 Rosemont diagnostic stratification.

Stratum	Criteria
Consistent with CP	1 major feature A + ≥ 3 minor features
	1 major feature A + major feature B
	2 major feature
Suggestive of CP	1 major feature A + < 3 minor features
	1 major feature B + ≥ 3 minor features
	≥ 5 minor features (any)
Indeterminate for CP	3 or 4 minor features major feature B alone or with < 3 minor features
Normal	≤ 2 minor features1

¹Excludes cysts, dilated main pancreatic duct, hyperechoic non-shadowing foci, dilated side branch. CP: Chronic pancreatitis.

Using EUS-FNA for diffuse CP, the negative predictive value increased to 100% against 75% for EUS, the specificity increased to 67% vs 60%, with higher concordance for severe disease than for mild CP[38]. Tru-Cut biopsy should not be recommended for non-focal CP because of complications[39], but its utility has been proved in autoimmune pancreatitis[31,40].

DIFFERENTIAL DIAGNOSIS

If focal hypoechoic lesion are found in the pancreatic parenchyma, the differential diagnosis includes primary or secondary pancreatic tumor, focal CP, and autoimmune pancreatitis. Several methods have been developed for this purpose.

Elastography

Elastography evaluates tissue strain resulting from compression and that strain is smaller in harder tissue than in softer tissue. Different tissue elasticity patterns are marked supplementary on the grey-color scale with different colors (blue for hard tissue and red for soft tissue). EUS elastography in CP shows a honeycomb aspect with predominantly hard strands, corresponding to fulfillment of four standard diagnostic criteria. The sensitivity, specificity, and accuracy were found to be 66%, 57% and 60%, respectively, and the method was considered useful in cases of equivocal EUS (three criteria or fewer)[41,42]. Further studies overcame the limitations of qualitative image analysis by means of digital image quantification, which helps to differentiate benign (normal pancreas and chronic pseudotumoral pancreatitis) from malignant lesions (pancreatic cancer and neuroendocrine tumors) with higher sensitivity, specificity, and accuracy (91.4%, 87.9% and 89.7%, respectively)[43]. Using a scoring system based on different color patterns in the images, the differentiation between benign and malignant pancreatic masses had sensitivity of 92.3% and specificity of 80%[44]. However, another study concluded that elastography did not allow complete delineation of the border of lesions greater than 35 mm in diameter or of lesions situated at some distance from the transducer, yielding poor sensitivity (41%), specificity (53%), and accuracy (45%) for predicting the nature of pancreatic focal lesions[45]. Because elastographic images are still difficult to obtain and interpret, although interobserver agreement is good (k = 0.725)[44], further improvement of the equipment with the possibility of quantification is expected. EUS elastography could have a special role in autoimmune pancreatitis, where the whole pancreas shows a typical, unique homogeneous stiffness, distinct from the circumscribed mass lesion in ductal adenocarcinoma[46].

Contrast-enhanced EUS

Ultrasound contrast agents increases the signal from the blood and improves the detectability of small vessels flow during ultrasound examinations. Before and after injection of Sonovue^® (Bracco), the focal pancreatitis shows no detectable vascularization or the vessels appear regular over a distance of at least 20 mm, with detection of both arterial and venous vessels in the contrast-enhanced phase[47] (Figure 3). Based on the perfusion characteristics of microvessels, contrast-enhanced US facilitates differential diagnosis between inflammatory lesions and ductal adenocarcinoma. The specificity of the discrimination between benign and malignant focal pancreatic lesions was found to be 93.3% using power Doppler contrast-enhanced EUS (CE-EUS) compared with 83.3% for conventional EUS[47]. The hypovascular aspect of lesions under power Doppler CE-EUS seemed highly sensitive and specific (91.1% and 93.3%, respectively) for adenocarcinoma[48]. During power Doppler CE-EUS examinations the ultrasound frequency returned to the transducer is the same with that transmitted, but the method is associated with artifacts resulting from turbulent flow (blooming and overpainting). The use of contrast agents is preferred using harmonic frequencies which result from non-linear and non-symmetrical oscillation of the microbubbles. This yields an image with complete “subtraction” of the tissue-derived signal, optimized by using a low mechanical index, which allows continuous real-time assessment of the microvascularization during contrast medium uptake. Harmonic CE-EUS shows an iso vascular homogeneous pattern of CP[49] or, in severe forms, a hypovascular pattern, due to extensive fibrosis[50] (Figure 4A). Our results confirmed that severe CP may be hypovascular on harmonic CE-EUS, and quantitative assessment of images can improve differentiation between adenocarcinomas and chronic pancreatitis (accuracy of 86%) (unpublished data), but, similar to elastography, cannot replace the use of EUS-FNA.

Figure 3 Mass resembling chronic pancreatitis. A: Conventional endoscopic ultrasonography (EUS). Hypoechoic inhomogeneous mass in the pancreatic head. Aorta and inferior caval vein are also seen; B: Contrast-enhanced harmonic-EUS. During the arterial phase (25 s after contrast injection) the abdominal aorta becomes hyperechoic and the mass is hypovascular compared with surrounding parenchyma.

Figure 4 Endoscopic ultrasonography-guided pseudocyst drainage. A: The cystostomy is seen as a hyperechoic parallel structure inside the hypoechoic well-delineated pseudocyst; B: Endoscopic view of a stent and a nasocystic drainage placed transgastric into a pseudocyst.

EUS-FNA of focal lesions

The EUS sensitivity for detection of suspected pancreatic mass in a parenchyma with CP modifications was 100%, but the positive predictive value of pancreatic malignancy in these situations was only 60%, because some malignant masses present internal or peripheral calcifications, similar to focal CP[51]. The sensitivity of EUS-FNA for malignancy in parenchymal masses with features of CP is only 54%-74%, compared with 89% when the surrounding parenchyma is normal[51-55]. However, in the event of high suspicion of malignancy with negative EUS-FNA, repeated FNA yields a positive diagnosis in 84% of cases, whereas half of the failures of first biopsies are attributed to the presence of CP[56]. Kras mutation and allele deletion of the microsatellite or of the tumor suppressors can be reliably detected in EUS-FNA samples from pancreatic masses, improving the diagnostic accuracy[57,58]. The search for codon-12 Kras mutation revealed no cases in patients with pseudotumor CP, in contrast to the adenocarcinoma group, although 6%-12% of patients with diffuse CP and PanIN lesions had presented Kras mutations in a previous meta-analysis[59,60].

EUS THERAPY

EUS-guided celiac block

One of the therapeutic uses of EUS in CP is celiac plexus blockade, i.e. temporary inhibition of the celiac plexus using a combination of local anesthesia and steroids, with the aim of reducing pain and improving the quality of life[61]. This guidance is preferred to CT-guided blockade because the details of the region are better appreciated and the side effects are fewer and less severe[62]. Frequently the celiac ganglia can be seen as a unique or concatenate hypoechoic structure, less well delineated, with some whitish strands inside[63].

Some issues regarding EUS-guided celiac block remain to be resolved. The indication is pain in CP, but some studies included pain accompanying moderate pancreatitis[64] or patients with pain that had not responded to other forms of treatment[65]. Another unclarified issue is the technique of injection (central or bilateral) and the quantity of steroid needed. The majority of studies used the bilateral injection technique, considered equal in safety to central injection, but the results of the two techniques concerning the alleviation of pain were close and contradictory[64,66], showing the need for a placebo-controlled trial[67]. Direct injection of triamcinolone within the celiac ganglia (13 patients) compared with alcohol injection (5 patients) yielded disappointing results in respect of pain alleviation for steroid use (38% vs 80%)[68]. A comparative study of results between the celiac region injection and celiac ganglia injection for EUS-guided celiac block is still lacking.

The question of cost-effectiveness remains unresolved. Some studies followed up the patients for only 1-4 wk[66,68]. The only study with an extended follow-up period showed duration of pain relief of up to 673 d. This raises the question of whether the natural course of the disease may have been responsible, because there were no data indicative of the level of severity of CP: duration of disease from onset of pain, presence of diabetes, or calcifications[64].

In many studies, the alleviation of pain varied from 55% to 70% with a short duration of follow-up[64-66,69]. Persistence of pain alleviation for as long as 24 wk was seen in no patients[65] or in only 10% of patients[69]. Two meta-analyses showed efficacy in managing chronic abdominal pain in 51.46%[70] and 59.45%[71] of patients respectively. The rate of major complications seemed very low (0.6%), being represented by retroperitoneal abscess[72].

EUS-guided drainage of pseudocysts

Therapeutic intervention in patients with chronic pancreatic pseudocysts is indicated when at least one complication is present (compression of large vessels, obstruction of duodenum, stomach, or common bile duct, infection, hemorrhage into pancreatic pseudocyst, pancreatico-pleural fistula) or when symptoms occur (satiety, pain, nausea or vomiting, upper gastrointestinal bleeding)[73,74]. Since 1996, several series of EUS-guided drainage have been reported, especially for collections without bulging onto the gut wall or with parietal vessels due to portal hypertension[75-77]. The main limitation is location of the fluid collection further than 1 to 1.5 cm from the gut wall[78-80] (Figure 4).

This method is preferred to surgical drainage, which is associated with a high rate of mortality and morbidity[81]. However, a non-randomized case-control study showed the same rates of treatment success, complications, and reinterventions for surgical and EUS-guided drainage, but with lower costs and shorter hospital stay for the EUS-guided procedure[82].

Conventional endoscopic drainage and EUS-guided drainage were compared in four papers. In a prospective non-randomized study the two approaches seemed equally safe and effective[83], but this was not confirmed in a second non-randomized study, where EUS represented a salvage method in the case of failure of conventional endoscopic drainage owing to non-bulging pseudocysts or location in the tail of the organ, but was a more time-consuming procedure[84]. The conclusion of this second study was that EUS should be reserved for pseudocysts located in the tail of the pancreas, because these are unlikely to cause luminal compression or are technically difficult to access. Also, EUS assessment would identify a tumor in 5% of pseudocysts[84]. A third randomized clinical trial showed a significantly better success rate for EUS- than for conventional endoscopic-guided drainage (100% vs 33%), despite the small number of patients, even after statistical adjustment for luminal compression[85]. A fourth randomized study confirmed also a significant advantage for EUS over conventional endoscopic drainage (94% vs 72%); both were considered first-line methods for treatment of bulging pseudocysts, but the authors recommended that EUS-guided drainage should be preferred for non-bulging pseudocysts[86].

Several aspects of EUS-guided drainage remain to be elucidated. First among these is the issue of the means used to create the communication between gut and pseudocyst. There are two major techniques for obtaining this communication: (1) balloon dilatation of a previous puncture site, with a 93%-100% success rate[83,84,87-89]; and (2) coagulation of the communication site by means of a cystostomy (success rate of 95% when two procedures per patient were performed[90] and 71%-82% with one procedure per patient[91,92]), a Giovannini needle (success rate of 94%[93,94], but only 84% after the first attempt[86]), or a needle-knife, with the same success rate as balloon dilatation but a higher perforation rate[88,89,95,96]. Larger comparative studies will be necessary to assess the best device with the highest success rate and the lowest complication rate. The prototype “transluminal balloon accessotome”, which combines a needle-knife and a dilating balloon, will probably allow easier drainage in one single step, reducing the exchange of accessories and simplifying the procedure[97]. Moreover, the use of the prototype three-layer puncture kit, which allows the simultaneous insertion of two guidewires at the initial puncture in one step, or the use of a larger working channel in the echo-endoscope, would allow safer and faster drainage[98]. Furthermore, the use of a forward-viewing echoendoscope seems promising for drainage of pseudocysts, even those inaccessible with a conventional therapeutic side-viewing EUS endoscope[99].

A further issue to be resolved is that of the morphological or biological factors that predict therapeutic success. Knowledge of such factors would facilitate selection of patients suitable for direct surgery. Moreover, to avoid pseudocyst relapse, described in 4%-17% of cases after 6-9 mo follow-up[94,96,100], communication with a secondary pancreatic duct, should be assessed very carefully.

EUS-guided drainage of main pancreatic duct

EUS-guided drainage of the MPD is a second-line procedure indicated when ERCP is unsuccessful owing to inability to cannulate the MPD (severe inflammation, previous surgery, postsurgical stricture) or difficult endotherapy (tight stenosis, large stone, MPD rupture, pancreas divisum). In practice, there are only few cases in which ERCP cannot be successfully performed by an experienced endoscopist, and recent studies suggests the superiority of surgery in managing pain. Thus, only a very small number of patients, namely those in whom ERCP fails and surgery cannot be performed safely, are good candidates for this procedure[101]. Using the transluminal approach or the transpapillary rendezvous approach, EUS-guided drainage of the MPD remains technically challenging because of difficulty in orienting the endoscope along the axis of the duct, difficult dilatation of the transmural tract due to pancreatic fibrosis, or the acute angle of the needle in relation to the MPD. Despite success rates of 68%-71%, the complication rates were important in all four series published (5%-43%); the complications included perforations, bleeding, pancreatitis, fever, and postprocedural pain[102-105]. EUS-guided drainage of the MPD should continue to be confined to tertiary care centers and very experienced endoscopists.

CONCLUSION

The diagnosis of CP is still accomplished using the standard scoring based on nine criteria each considered as having the same value. For diagnosis of any CP, at least three or four of these criteria must be present, but for diagnosis of severe CP more than six criteria must be fulfilled. The more restrictive Rosemont classification aims to standardize the criteria and assigns different values to different features, but requires further validation. EUS-FNA is less advisable for diagnosis of diffuse CP due to the possible side effects. Elastography and contrast-enhanced EUS are orientation in differentiating focal pancreatic mass, but they cannot replace EUS-FNA. The utility of EUS-guided celiac block for painful CP is still a matter of debate with regard to best technique and the indications. EUS-guided drainage of pseudocysts is preferred especially in non-bulging pseudocysts or presence of portal hypertension. EUS-guided drainage of the MPD should be reserved for cases of unsuccessful ERCP caused by difficult cannulation of the papilla or difficult endotherapy. It should be performed only by highly skilled endoscopists, due to the high risk of complications.