Diagnosis and treatment of autoimmune pancreatitis types 1 and 2

Abstract

Background

Autoimmune pancreatitis (AIP) is characterized by diffuse or focal swelling of the pancreas. AIP has been divided into types 1 and 2. The aim of the study was to evaluate and compare the clinicopathological characteristics, therapy and outcome of patients with AIP.

Methods

The medical records of patients diagnosed with AIP between January 2003 and July 2011 were reviewed. Characteristics of patients with AIP types 1 and 2 were compared with those of patients with pancreatic ductal adenocarcinoma (PDAC).

Results

AIP was classified as type 1 in 40 patients and type 2 in 32 according to the HISORt (Histology, Imaging, Serology, Other organ involvement, Response to therapy) criteria. Patients with histologically confirmed AIP type 2 were younger than those with type 1 (P = 0·005). Some 30 of 32 patients with AIP type 2 were found to have a localized tumour-like pancreatic mass and underwent pancreatectomy, compared with only 16 of 40 with type 1 (P < 0·001). Three of 25 patients with AIP type 2 presented with raised serum levels of IgG4 compared with 21 of 38 with type 1 (P < 0·001). There was no difference in symptoms and involvement of other organs between AIP types 1 and 2. Presentation with weight loss was more common among patients with PDAC than those with AIP, but there was no difference in pain or jaundice between the groups. Raised serum carbohydrate antigen 19-9 levels were more prevalent in patients with PDAC.

Conclusion

Patients with AIP type 2 frequently present with abdominal pain and a tumour-like mass. Differentiating AIP from PDAC is difficult, so making the clinical decision regarding operative versus conservative management is challenging.

Introduction

During the past decade, recognition of autoimmune pancreatitis (AIP) as a distinct clinical entity has increased worldwide^1,2. AIP accounts for approximately 5 per cent of all patients with chronic pancreatitis³. The aetiology and pathogenesis remain poorly understood⁴. Patients with AIP frequently present with diffuse or focal swelling of the pancreas with or without jaundice owing to extrahepatic cholestasis⁵. Even with use of modern thin-slice abdominal imaging the differential diagnosis between AIP and a malignant pancreaticobiliary tumour remains challenging^6,7. Consequently, affected patients are frequently referred for surgery and undergo pancreatic resection to exclude definitely the risk of harbouring a malignant pancreatic tumour⁸.

During the international AIP consensus meeting in 2009⁹, the need to differentiate between two subtypes of AIP was outlined^10–13. AIP types 1 and 2 differ with regard to treatment response and relapse rate. Type 1 fits the classical criteria of AIP and is associated with a histological pattern of lymphoplasmacytic sclerosing pancreatitis^1,14–16. Affected patients usually present with abdominal pain, weight loss, anorexia and jaundice. There is frequently diffuse swelling of the pancreas, often associated with an irregular narrowing of the pancreatic duct¹⁷. Some 75 per cent of patients with AIP type 1 present with raised serum levels of IgG4¹⁸. AIP type 1 is frequently associated with other autoimmune-related diseases^6,19–20.

AIP type 2 is characterized by a so-called idiopathic duct-centric pancreatitis and pathognomonic ‘granulocyte epithelial lesion (GEL)-positive pancreatitis’^21,22. First described by Ectors and colleagues²³, GELs are characterized by focal detachment, disruption and destruction of the epithelium of interlobular ducts²⁴. Although the histomorphology of AIP type 2 has been described in various reports, little is known about the clinical characteristics and outcome of these patients¹¹. The diagnosis of AIP type 1 can usually be made on clinical grounds, by imaging²⁵ and laboratory tests^1,19,26, whereas histological proof is mandatory for the definite diagnosis of AIP type 2^12,27–28.

The aim of the study was to evaluate the clinicopathological characteristics, therapy and outcome of patients with AIP types 1 and 2.

Methods

All patients seen at the Pancreas Centre of the Department of Surgery, University of Heidelberg, who fulfilled the HISORt (Histology, Imaging, Serology, Other organ involvement, Response to therapy) criteria²⁹ for AIP between January 2003 and July 2011 were eligible for the study. Following the HISORt classification system, AIP was diagnosed when patients met at least one of the following criteria: histologically confirmed diagnosis of AIP following surgical resection or biopsy (HISORt diagnostic criteria group A); typical radiological features (such as diffusely enlarged sausage-shaped gland with irregular narrowing of the main pancreatic duct) and associated serological changes (HISORt diagnostic criteria group B); and typical radiological features of AIP and an imaging-confirmed response to steroid therapy (HISORt diagnostic criteria group C). Patients with AIP type 2 were diagnosed exclusively based on histology. If histology was not available, but the patient met the HISORt criteria, the patient was assigned to the AIP type 1 cohort¹⁹.

Patients with known risk factors for chronic pancreatitis, such as alcohol consumption exceeding 60 g/day, pancreas divisum, history of necrotizing pancreatitis or abdominal trauma, were excluded from the study.

A standard record form, including details of age and sex, symptoms, other organ involvement, abdominal imaging, blood tests, treatment and quality of life, was filled out by the physician at each visit. Any specific information relevant to the study (recurrent disease, surgical procedures, complications) were documented. Evaluation of the patients' medical history focused on pancreaticobiliary diseases, previous pancreatic surgery, autoimmune-related diseases including sclerosing cholangitis, sclerosing sialoadenitis, involvement of lacrimal glands or kidneys, retroperitoneal fibrosis, and allergic manifestations such as allergic rhinitis, atopic dermatitis or bronchial asthma³⁰. This study and the questionnaire used were approved by the local ethics committee of the University of Heidelberg.

Imaging studies

All patients underwent either multidetector computed tomography (MDCT), magnetic resonance imaging (MRI) and/or magnetic resonance cholangiopancreatography (MRCP). Endoscopic retrograde pancreatography and endoscopic ultrasonography were considered for evaluation of pancreatic morphology when available. Radiological features characteristic of AIP included diffusely enlarged pancreas with loss of normal lobularity (sausage-like appearance) (Fig. 1), delayed and low-density ‘rim’ enhancement, and a dilated main pancreatic duct together with focal or diffuse narrowing²⁵. ‘Tumour-like lesion’ was used to describe a focal mass-forming lesion resembling a pancreatic malignancy (Fig. 2). Thin-slice imaging was also used to exclude potential metastatic disease and to assess whether there were abdominal extrapancreatic autoimmune-related lesions.

Fig. 1

Typical sausage-shaped pancreas of a 51-year-old man with autoimmune pancreatitis type 1. Multidetector computed tomography in the portal-venous phase shows diffuse swelling of the entire pancreas (arrows)

Fig. 2

Multidetector computed tomography with intravenous contrast showing tumorous lesion in the pancreatic head of a patient with autoimmune pancreatitis type 2: a axial plane (note endoprosthesis in distal part of bile duct) and b coronal view

Blood tests

Serum concentrations of IgG and IgG4 were determined at the time of diagnosis by automated immunonephelometry. In all patients who underwent primary surgery for suspected pancreatic cancer, IgG levels were determined as soon as the final histological diagnosis of AIP was available. Levels of IgG of at least 16·0 g/l, and IgG4 of 140 g/l or more, were considered raised³¹. The following autoantibodies were used as markers for autoimmune disease; levels were considered to be raised when values were above the upper limit of normal (in parentheses): antinuclear antibody (ANA; more than 1 : 80), rheumatoid factor (RF; over 25 units/ml), antineutrophil cytoplasmic antibodies (ANCA; more than 1 : 10), male germ cell-associated kinase antibody (MAK; over 60 units/ml), thyroid-stimulating hormone receptor antibodies (TRAK; over 1·00 units/ml), lactoferrin antibody (antibody ratio less than 10), carbonic anhydrase 1 antibodies (positive), carbonic anhydrase 2 antibodies (positive), extractable nuclear antigens (ENA; more than 6·3 units/l) and soluble interleukin 2 receptor (sCD25; over 900 units/ml). ANA, ANCA, MAK, TRAK, ENA, RF and sCD25 levels were determined using enzyme-linked immunosorbent assays.

Histopathology

Haematoxylin and eosin staining, and immunohistochemistry for IgG4-positive plasma cells was carried out routinely in all resected specimens and biopsies. Diagnostic criteria for AIP included periductal lymphoplasmocytic infiltrate, periductal concentric fibrosis (Fig. 3a) and detection of vasculitis (Fig. 3c)¹³. AIP type 2 was diagnosed when an idiopathic duct-centric pancreatitis with typical GELs was present (Fig. 3b) in combination with absent or rare IgG4-positive plasma cells (10 or fewer per high-power field (HPF)). AIP type 1 was characterized by lymphoplasmacytic sclerosing pancreatitis including dense fibrosis with rich lymphoplasmacytic infiltration (‘storiform fibrosis’), and obliterative phlebitis (Fig. 3c). In addition, numerous IgG4-positive plasma cells (more than 10/HPF) were identified in AIP type 1 (Fig. 3d)²⁷. Histopathological diagnoses were made in accordance with the 2011 international consensus diagnostic criteria for autoimmune pancreatitis²⁸.

Fig. 3

Histological features of autoimmune pancreatitis: a periductal concentric fibrosis (arrow) and periductal infiltrates of lymphocytes and plasma cells (dotted arrow) (haematoxylin and eosin stain, original magnification ×18); b granulocytic intraepithelial lesions (arrow) in type 2 autoimmune pancreatitis (haematoxylin and eosin stain, original magnification ×72); c venulitis (arrow) (haematoxylin and eosin stain, original magnification ×72); d increased number of immunohistochemically IgG4-positive plasma cells (stained red) in autoimmune pancreatitis type 1 (methylene blue counterstain, original magnification ×72)

Follow-up

All patients, whether managed conservatively or by surgery, were seen in the outpatient clinic after 3 months. Thereafter the interval for future visits was determined on an individual basis depending on clinical symptoms or other complaints. The date of last follow-up was July 2011. Follow-up data were obtained in the outpatient clinic or from the patient's last visit to the general physician. Evaluation included the current status of the disease, as well as the occurrence or relapse of pancreatic or extrapancreatic manifestations.

Comparison of autoimmune pancreatitis with pancreatic ductal adenocarcinoma

All patients with AIP type 1 or 2 who presented with a localized lesion suspicious for malignancy were matched with controls with pancreatic ductal adenocarcinoma (PDAC). The following parameters were used to match one patient with AIP (case) with two patients with PDAC (controls): age (± 5 years), sex and site of disease. All controls were identified from the institutional database of pancreatic cancer.

Statistical analysis

The records of all patients diagnosed with AIP were entered into a database. Continuous data are presented as median (range), unless indicated otherwise. The distribution of age was compared between groups using the Mann–Whitney U test. Fisher's exact test was used for analysis of categorical variables. All tests were two-sided and statistical significance was assumed at the 5 per cent level. Statistical evaluation was performed using the SAS® version 9.1 for Windows® (SAS Institute, Cary, North Carolina, USA).

Results

Some 72 patients (52 men, 20 women) with AIP were included in the study. None of the patients underwent core biopsy for preoperative diagnosis. A total of 47 patients had a pancreatic resection. AIP type 1 was diagnosed in 40 patients, and in 25 the diagnosis was established by clinical presentation, imaging and serology (HISORt diagnostic criteria group B and C) (Fig. 1). In 15 patients, histopathological examination of the resection specimen was performed (HISORt diagnostic criteria group A). There were 32 patients with AIP type 2, all diagnosed after pancreatectomy.

Characteristics of the patients with AIP types 1 and 2 are summarized in Table 1. Patients presented with abdominal pain and a medical history of acute or recurrent acute pancreatitis. One-third were jaundiced and nine of the 32 patients with AIP type 2 had associated autoimmune diseases. Patients with AIP type 2 were younger than patients with type 1 (P = 0·267), and more often showed a localized pancreatic mass on imaging (30 of 32 versus 16 of 40; P < 0·001). Patients with AIP type 2 had a lower prevalence of raised serum levels of IgG (2 of 24 versus 8 of 38; P = 0·048) and IgG4 (3 of 25 versus 21 of 38; P < 0·001).

Table 1

Characteristics of patients with autoimmune pancreatitis types 1 and 2

	AIP type 1 (n = 40)	AIP type 2 (n = 32)	P‡
Age (years)*	52 (18–74)	44·5 (20–75)	0·267§
Sex ratio (M: F)	29 : 11	23 : 9	> 0·999
Clinical presentation
Abdominal pain	32	25	> 0·999
Jaundice	13	12	0·804
Diarrhoea/steatorrhoea	14	8	0·444
Acute or recurrent acute pancreatitis	20	12	0·344
Weight loss	9	7	> 0·999
Site of disease
Head/body	12	27	< 0·001
Tail	1	4	0·164
Entire gland	27	1	< 0·001
Imaging
Diffuse pancreatic swelling	26	1	< 0·001
Tumour-like pancreatic mass (suspicious for malignancy)	16	30	< 0·001
Serology
Raised IgG (≥16 g/l)	8 of 38	2 of 24	0·048
Raised IgG4 (≥140 g/l)	21 of 38	3 of 25	< 0·001
Raised ANA	12 of 38	8 of 25	0·792
Other raised autoimmune antibodies	20 of 38	10 of 25	0·150
Autoimmune-related diseases	13	9	0·799
Psoriasis	0	2	0·194
Rheumatism	5	1	0·217
Autoimmune gastritis	3	2	> 0·999
Hashimoto's thyroiditis	1	2	0·581
Hepatitis, SLE	2	2	> 0·999
Sjögren's syndrome	3	1	0·624
Inflammatory bowel disease	0	2	0·194
Initial therapy†
Pancreatic resection	13	31	< 0·001
Corticosteroids	16	1	< 0·001
Best supportive care	11	0	< 0·001

*

Values are median (range).

†

First 2 weeks after diagnosis. AIP, autoimmune pancreatitis; ANA, antinuclear antibody; SLE, systemic lupus erythematosus.

‡

Fisher's exact test, except

§

Mann–Whitney U test.

Comparing only histologically confirmed AIP type 1 (15 patients) with type 2 (32 patients), a higher proportion of patients with AIP type 1 had raised levels of IgG4 (8 of 13 versus 3 of 25; P = 0·002). Patients with AIP type 2 were significantly younger (P = 0·005), but all other clinical, laboratory and imaging data were comparable between types 1 and 2.

Treatment of autoimmune pancreatitis

Most patients with AIP type 1 were treated conservatively in the first 3 weeks after diagnosis with corticosteroids or best supportive care, whereas the majority with AIP type 2 underwent surgical resection (Table 1). Fifteen of the 40 patients with AIP type 1 eventually underwent pancreatectomy. The indications for surgical treatment, type of operation and outcome within 28 days are summarized in Table 2.

Table 2

Indications for surgery, procedures and outcomes

	AIP type 1 (n = 40)	AIP type 2 (n = 32)
Indication for surgical treatment
Pancreatic resection to exclude malignancy (tumour-like mass)	9	24
Pain following conservative management of pancreatitis	2	3
Unsuccessful steroid therapy	0	1
Pancreatic duct stenosis*	0	2
Cystic pancreatic lesion suspicious for IPMN	1	1
Total pancreatectomy for suspected total gland disease	0	1
Cholestasis suspicious for malignant obstructive process	3	0
Operative procedure
Pancreaticoduodenectomy	11	23
Duodenum-preserving pancreatic head resection	2	4
Total pancreatectomy	1	1
Distal pancreatectomy	1	4
Outcomes
Death	0	0
Mean hospital stay (days)	9·5	10
Reoperation (postop. bleeding)	0	1
Delayed gastric emptying	3	4
Pancreatic fistula	0	1
Lymphatic fistula	0	1
Cholangitis	0	1
Intra-abdominal abscess	1	0
Surgical-site infection	1	0
Pneumonia	0	1
Urinary tract infection	1	0

*

Pancreatic resection for pancreatic duct stenosis following previous pancreatic surgery (duodenum-preserving pancreatic head resection and left pancreatic resection). AIP, autoimmune pancreatitis; IPMN, intraductal papillary mucinous neoplasm.

Clinical course after surgery for autoimmune pancreatitis

Follow-up was available for at least 6 months after surgery for all but one patient with AIP type 2. After operation, all patients with AIP type 2 showed a decrease in abdominal pain. Typical disease relapse was characterized by recurrent episodes of pain and pancreatitis. Six patients with AIP type 2 received corticosteroid therapy. Five patients received corticosteroids after operation, including four patients with recurrent disease. Recurrence developed within a median of 34 (3–50) months. One patient received corticosteroids 4 weeks after surgery as prophylactic therapy and one before surgery as initial treatment.

Among the patients with AIP type 1 who underwent pancreatic resection, nine complained of abdominal pain before the operation. Four of these patients had raised serum levels of carbohydrate antigen (CA) 19-9 (at least 38 units/ml) and/or carcinoembryonic antigen (5 µg/l or more) in preoperative investigations. Seven of 15 patients had a raised total serum bilirubin level (1·5 mg/dl or more). After surgical resection, seven of nine patients experienced relief of pain, two reported unchanged pain intensity after surgery, and one experienced new-onset mild abdominal pain.

Comparison between autoimmune pancreatitis and pancreatic ductal adenocarcinoma

Table 3 shows a comparison of patients with AIP and matched patients with PDAC. Seven of those with AIP aged between 15 and 32 years were matched with patients with PDAC up to 20 years older. One 48-year-old man presented with a mass-forming tumour in the tail of the pancreas that was suspected to be a neoplastic tumour, but had features suggestive of AIP. Owing to his relatively young age and preference, he received conservative therapy with corticosteroids for 2 weeks, which was successful. Weight loss was more common among patients with PDAC than those with AIP. Raised serum CA19-9 levels were more prevalent among patients with PDAC. There was no difference in pain or jaundice.

Table 3

Comparison between autoimmune pancreatitis and pancreatic ductal adenocarcinoma for each subtype of autoimmune pancreatitis

	AIP type 1 (n = 16)	PDAC (n = 32)	P	AIP type 2 (n = 30)	PDAC (n = 60)	P†
Age (years)*	63 (24–71)	64 (39–71)	0·956	45 (16–79)	46 (33–79)	0·327‡
Sex ratio (M: F)	16 : 0	32 : 0	–	21 : 9	42 : 18	1·000
Site of disease			1·000			1·000
Head/body	14 (88)	28 (88)		27 (90)	54 (90)
Tail	2 (12)	4 (12)		3 (10)	6 (10)
Pain	10 (63)	15 (47)	0·369	23 (77)	40 (67)	0·465
Jaundice	11 (69)	21 (66)	1·000	13 (43)	37 (62)	0·114
Weight loss	4 (25)	24 (75)	0·002	6 (20)	40 (67)	< 0·001
Acute pancreatitis	5 (31)	2 (6)	0·033	11 (37)	8 (13)	0·015
Tumour markers
CA19-9 > 37 units/ml	2 (13)	24 (75)	< 0·001	5 (17)	47 (78)	< 0·001
CEA > 2·5 µg/l	5 of 15 (33)	19 (59)	0·125	5 (17)	23 of 59 (39)	0·052

Values in parentheses are percentages unless indicated otherwise;

*

values are median (range). AIP, autoimmune pancreatitis; PDAC, pancreatic ductal adenocarcinoma; CA, carbohydrate antigen; CEA, carcinoembryonic antigen.

†

Fisher's exact test, except

‡

Mann–Whitney U test.

Discussion

Although there have been consensus conferences in Asia and the USA to establish an exact definition for the diagnosis of AIP, no clear characterization and classification of the different subtypes is yet available. To determine the correct classification for diagnosis of AIP, the Japanese criteria³², HISORt criteria²⁹, consensus Asian criteria²⁶, Korean criteria³³ and the M-ANNHEIM classification³⁴ were reviewed. Although all classifications have overlapping criteria, there are important differences. According to the Japanese diagnostic criteria, typical imaging is mandatory for the diagnosis of AIP. In the present study of predominantly surgical patients with a histological diagnosis, preoperative imaging was not specific for AIP in many patients. These patients would not have been diagnosed with AIP using the Japanese criteria. It has been shown that the HISORt criteria have the highest sensitivity (92 per cent) for the diagnosis of AIP³⁵. These criteria are in accordance with the 2011 international consensus diagnostic criteria for AIP²⁸ and seemed most appropriate for the present study.

The most feared scenario is PDAC misdiagnosed as AIP. Recent advances in MDCT, MRI and/or MRCP imaging techniques have led to high-quality image acquisition that has improved the ability confidently to recognize typical features of AIP¹⁷. However, it remains a challenge to differentiate between AIP and pancreatic cancer, particularly in patients with AIP type 2 who present with a suspected malignant lesion. In these patients, late enhancement seems to be a key feature and is best detected with MRI. With regard to symptoms, patients with PDAC more frequently presented with weight loss than those with AIP in the present study. However, no weight loss occurs in one in three patients with pancreatic cancer, particularly in the earlier stages. Thus, this clinical variable does not seem to be helpful in discriminating between PDAC and AIP in individual patients. Likewise, in patients with a lesion suspicious for AIP and normal serum tumour marker levels (particularly CA19-9), PDAC cannot be excluded reliably, because a raised level of CA19-9 is found in only 75 per cent of patients with PDAC³⁶.

It has been reported that the serum IgG4 level is raised in 94 per cent of patients with AIP³⁷. In the present study, only 55 per cent of patients (21 of 38) with AIP type 1 and 12 per cent (3 of 25) with type 2 had raised IgG4 levels, in accordance with a recent report³⁸. IgG4 levels may be raised in up to 10 per cent of subjects without AIP, which further complicates the differential diagnosis of pancreatic cancer³⁹. Therefore, IgG4 is not a good marker with which to identify patients with AIP. Surgical resection is warranted if there is any doubt about the diagnosis and a tumour-like pancreatic mass is present. Alternatively, corticosteroid therapy can be given for 2 weeks to patients with typical AIP morphology on MRI, followed by further MRI to evaluate the response to steroids. However, the treatment response always must be evaluated critically, and it has to be kept in mind that serum IgG4 levels may ‘improve’ with steroid therapy even though the underlying disease is pancreatic cancer⁶.

The present study has confirmed that histology should be the standard, not only for the diagnosis of AIP, but also to differentiate between the two subtypes. In contrast to current literature, there were few clinical differences between AIP types 1 and 2 in the present cohort. High serum levels of IgG4 were found primarily in patients with AIP type 1, but three of 25 patients with AIP type 2 also had raised serum levels of IgG4. In contrast to Park and colleagues¹¹, the present authors conclude that IgG4 serum levels are helpful neither in diagnosing AIP nor in differentiating between types 1 and 2.

The present study has confirmed that patients with AIP type 2 are on average 8 years younger than those with AIP type 1, as shown in an international multicentre study¹⁰. In accordance with the report of Sah and colleagues¹² and the Honolulu consensus document²⁷, there were no differences in sex distribution between the two AIP subtypes. Nor was there a difference in extrapancreatic manifestations between AIP types 1 and 2. Therefore, this study cannot confirm the widely held belief that AIP type 1 is a systemic IgG4-related disease, whereas type 2 is exclusively confined to the pancreas²⁰.

Beside IgG4 serum levels, there seems to be a difference between AIP types 1 and 2 with regard to morphology and location of the inflammatory mass. In the present study, patients with AIP type 1 more frequently showed diffuse swelling of the entire pancreas, whereas a tumour-like mass mimicking pancreatic cancer was found in the majority of patients with AIP type 2. Thus, in contrast to Maire and colleagues⁴⁰, who reported that surgery was performed more often in AIP type 1, the present results indicate that AIP type 1 can more frequently be diagnosed without histology and surgical resection, because this type is usually associated with typical features of AIP according to the Asian or HISORt criteria. In contrast, in patients with AIP type 2 the indication for operative versus conservative therapeutic management remains challenging.

Disclosure

The authors declare no conflict of interest.

Snapshot quiz 14/13