Abstract
Background
2017 International and 2018 European guidelines are the most recent guidelines for intraductal papillary mucinous neoplasms management.
Aim
to evaluate the diagnostic accuracy of these guidelines in identifying malignant IPMN.
Methods
data from resected patients with IPMN were collected in two referral centers. Features of risk associated to cancerous degeneration described in International and European guidelines were retrospectively applied. Sensitivity, specificity, positive and negative predictive value in detecting malignant disease were calculated.
Results
the study includes 627 resected patients. European guidelines suggest resection in any patient with at least one feature of moderate-risk. International guidelines suggest that patients with moderate-risk features undergo endoscopic ultrasound before surgery. European guidelines had a higher sensitivity (99.2% vs. 83%) but a lower positive predictive value (59.5% vs. 65.8%) and Specificity (2% vs. 37.5%). European guidelines detected almost all malignancies, but 40% of resected patients had low-grade dysplasia. 297 patients underwent endoscopic ultrasound before surgery. 31/116 (26.7%) tumors radiologically classified as “worrisome features” were reclassified as “high-risk stigmata” by endoscopic ultrasound and 24/31 were malignant IPMN.
Conclusions
European and International guidelines have a relatively low diagnostic accuracy, being European guidelines more aggressive. Endoscopic ultrasound can improve guidelines accuracy in patients with moderate-risk features.
Keywords
1. Introduction
The management of intraductal papillary mucinous neoplasms (IPMNs) has significantly evolved over time. Some twenty years ago surgery was the treatment of choice for all IPMNs [
[1]
,[2]
. Subsequently, IPMNs became a common finding in clinical practice, and it was evident that a high number of small, incidentally detected branch-duct IPMNs were benign lesions, while the risk of malignancy was much higher in main-duct and mixed IPMNs [[3]
,[4]
. These observations were included in the first International guidelines issued in 2006, that recommended surgical resection for patients with main duct IPMNs and non-operative management for asymptomatic branch duct IPMNs less than 30 mm in size lacking solid components [[5]
]. Since 2006, several guidelines were published, and most of these had a low positive predictive value (PPV) associated with a relatively high negative predictive value (NPV) for malignancy [5
, 6
, 7
, 8
, 9
, 10
]. These data reflect the difficulty to select appropriate candidates for surgical resection, as low-grade dysplasia is often found in resected IPMNs. This result can be considered an “un-necessary treatment” or even an “overtreatment”, especially for elderly patients whose life expectancy is probably not affected by an IPMN [[11]
]. This eventuality should be carefully balanced with the risk of missing IPMNs harboring invasive cancer (IC) or high-grade dysplasia (HGD). International (IG) and European guidelines (EG) have been updated in 2017 and 2018, respectively, in order to improve the identification of high-risk IPMNs [[12]
,[13]
.The aim of the present study is to evaluate the diagnostic accuracy of IG and EG in identifying malignant IPMNs by retrospectively applying the guidelines to a large cohort of resected IPMNs.
2. Methods
This retrospective study was carried out at the Division of Pancreatic Surgery of San Raffaele Scientific Institute, Milan, Italy and at HPB Disease Unit of Karolinska Hospital in Stockholm, Sweden. The study was approved by San Raffaele Scientific Institute Ethical Committee (protocol EMI-IPMN GR-2018-12,366,897) on 16/05/2019. The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a priori approval by the institution's human research committee. Due to its retrospective design, specific informed consent was not required since patients signed a general consent allowing clinicians to use data for clinical research. Patients with histologically confirmed IPMNs who underwent pancreatic resection between January 2009 and March 2018 were identified from prospective databases maintained at the two Institutions.
Demographics, symptoms, clinical history, diagnostic work-up (including laboratory, radiologic and endoscopic data), type of surgery, and pathology data were prospectively collected and then retrospectively analyzed.
In both Institutions the diagnosis was based on preoperative imaging and endoscopic studies, with the use of magnetic resonance imaging (MRI), magnetic resonance cholangiopancreatography (MRCP), multiple detector computed tomography (MDCT) and endoscopic ultrasound with fine needle aspiration (EUS+FNA). MRI with MRCP was considered the gold standard for the collection of imaging data; EUS was considered as a “second level” examination [
[12]
]. All patients were preoperatively discussed in conference [[14]
].High-risk stigmata (HRS) and worrisome features (WF) of IG and absolute indications (AI) and relative indications (RI) of EG were retrospectively applied [
[12]
,[13]
. HRS include jaundice, solid mass, main pancreatic duct (MPD) ≥10 mm, enhancing nodules ≥5 mm and cytology positive for high-grade dysplasia or adenocarcinoma. WF include cyst size ≥30 mm, MPD size of 5–9 mm, pancreatitis, enhancing mural nodule <5 mm, thickened/enhancing cyst walls, abrupt change in MPD caliber, lymphadenopathy, serum levels of CA 19.9 >37 U/mL and cyst growth rate >5 mm/2 years. AI include jaundice, MPD ≥10 mm, enhancing nodules ≥5 mm, solid mass and cytology positive for high-grade dysplasia or adenocarcinoma. RI include cyst size ≥40 mm, MPD size of 5–9.9 mm, pancreatitis, enhancing mural nodule <5 mm, serum levels of CA-19.9 >37 U/mL, new onset of diabetes and cyst growth rate >5 mm/year. Even though not clearly mentioned in the 2017 IG the “solid mass” criterion was considered among HRS, as it was an obvious indication for surgery since solid mass may represent the advanced form of an invasive IPMN [[15]
,[16]
. Patients with one or more HRS or AI were included only in these categories although they simultaneously had WFs or RIs. Histologic examination of the surgical specimen was based on the 2010 WHO classification system [[15]
]. IPMNs were histologically classified as low-grade dysplasia, moderate dysplasia, high-grade dysplasia and invasive carcinoma. IPMNs originally classified as moderate dysplasia were then re-classified as low-grade as suggested by the Baltimore consensus meeting [[16]
]. The presence of low-grade dysplasia defined an IPMN as “benign”, while IPMNs with high-grade dysplasia and invasive carcinomas were considered “malignant”.The primary aim of this work was to evaluate positive and negative predictive value (PPV, NPV), sensitivity (Sn), specificity (Sp) and accuracy of 2018 EG and 2017 IG in detecting malignancy.
Considering the different approach of the two guidelines for the management of fit patients with RI/WF (immediate surgery according to EG versus EUS eventually followed by surgery according to IG), secondary endpoints of the study were (i) to evaluate the role of EUS in the setting of patients with WF and (ii) to retrospectively apply IG versus EG management strategies to the study cohort.
Patients who met the absolute criteria for resection of the two guidelines and had a histological diagnosis of benign IPMN were considered as potentially “overtreated” - their surgery was not mandatory, especially in the setting of elderly people. Patients that didn't meet the straightforward surgery criteria and had a final diagnosis of malignancy were considered as potentially undertreated, since non-operative management could have been considered in this setting.
2.1 Statistical analysis
The performance of IG and EG was evaluated by calculating sensitivity, specificity, PPV, NPV and accuracy, with 95% confidence intervals, in predicting malignant IPMNs. Continuous variables are presented as the mean ± standard deviation (SD). Categorical variables are presented as numbers and percentages. Chi-square test was used to compare rates of malignancy between groups. Data were analyzed by means of SPSS version 20 (SPSS, Inc., Chicago, Il).
3. Results
3.1 Patients’ characteristics
In the study period, 627 patients, (327 males and 300 females; mean age: 68 ± 9 years) underwent pancreatic resection for a histologically confirmed IPMN. Table 1 shows the clinico-pathological characteristics of the cohort. The rate of MD/mixed-IPMNs and of BD-IPMNs were 83% and 17%, respectively. The rate of malignancy was 66% for MD/mixed-IPMNs compared with 25% in BD-PMNs (p <0.01). Table 2 shows the prevalence of AI and RI (2018 EG) and of WF and HRS (2017 IG) in the study population. Considering 2018 EG, 246 (39%) patients had RI and 373 (59%) AI. Applying 2017 IG, 248 (40%) patients had WF and 373 (59%) HRS. Five patients did not carry any HRS/WF and eight did not present AI/RI. These patients underwent surgery because of a family history of pancreatic cancer (n = 3) or because of personal choice (n = 5)
Based on the Baltimore classification
Table 1Clinico-pathological features of the 627 patients included in the study.
| Variable | Number of patients (%) | |
|---|---|---|
| Age (yrs.), mean±SD | 68.1 ± 9.4 | |
| BMI, mean±SD | 24.9 ± 4.3 | |
| Gender | Male Female | 327 (52%) 300 (48%) |
| Smoking status | No smoker Current smoker Former smoker | 403 (64%) 126 (20%) 98 (16%) |
| Diabetes | Yes No | 526 (84%) 101 (16%) |
| Family history of PDAC | No Yes | 609 (97%) 18 (3%) |
| IPMN site | Head Body/tail Diffused | 269 (43%) 165 (26%) 193 (31%) |
| IPMN type | BD-IPMN Benign Malignant • High grade dysplasia • Invasive carcinoma | 105 (17%) 79 (75%) 26 (25%) 7 (7%) 19 (18%) |
| MD-IPMN Benign Malignant • High grade dysplasia • Invasive carcinoma | 522 (83%) 177 (34%) 345 (66%) 118 (23%) 227 (43%) |
SD= Standard deviation, BMI= Body mass index, DM = diabetes mellitus type I or type II, PDAC= Pancreatic ductal adenocarcinoma; IPMN: intraductal papillary mucinous neoplasm; BD-IPMN: branch duct intraductal mucinous neoplasm; MD-IPMN: main duct- intraductal papillary mucinous neoplasm.
[15]
“benign IPMN” included low-grade dysplasia and intermediate grade dysplasia, while “malignant IPMN” included high-grade dysplasia and invasive cancer.Table 2Prevalence of absolute and relative indications and of high-risk stigmata and worrisome features in the cohort and according to 2017 International Guidelines and 2018 European Guidelines, and their correlation with benign and malignant IPMNs.
| IPMN characteristics | n (%) 619* | International Guidelines | European Guidelines | Benign IPMNs (n = 256) | Malignant IPMNs (n = 371) | ||
|---|---|---|---|---|---|---|---|
| HRS (n = 373) | WF (n = 248) | AI (n = 373) | RI (n = 246) | ||||
| Cyst ≥40mm | 177 (28%) | 84 (34%) | 90 (51%) | 87 (49%) | |||
| Enhancing nodule < 5mm | 46 (7%) | 20 (8%) | 20 (8%) | 14 (30%) | 32 (70%) | ||
| Cyst growth > 5 mm/yr | 42 (7%) | 32 (13%) | 24 (57%) | 18 (43%) | |||
| Increased serum CA-19.9 | 337 (54%) | 141 (57%) | 141 (57%) | 124 (37%) | 213 (63%) | ||
| Pancreatitis | 82 (13%) | 46(19%) | 46 (19%) | 43 (52%) | 39 (48%) | ||
| New onset of diabetes | 47 (7%) | 12 (5%) | 12 (26%) | 35 (74%) | |||
| Cyst ≥30mm | 222 (35%) | 111(45%) | 114 (51%) | 108 (49%) | |||
| MPD 5–9mm | 263 (42%) | 144 (58%) | 144 (59%) | 109 (41%) | 154 (59%) | ||
| Thickened cyst walls | 71 (11%) | 31 (13%) | 33 (46%) | 38 (54%) | |||
| Cyst growth >5 mm/2yrs | 56 (9%) | 38 (15%) | 31 (55%) | 25 (45%) | |||
| Abrupt change in MPD | 22 (35%) | 5 (2%) | 8 (36%) | 14 (64%) | |||
| Lymphadenopathy | 0 | 0 | – | – | |||
| Jaundice | 127 (20%) | 127 (34%) | 127 (34%) | 20 (16%) | 107 (84%) | ||
| MPD ≥10mm | 200 (32%) | 200 (54%) | 200 (54%) | 53 (27%) | 147 (73%) | ||
| Enhancing nodule ≥ 5 mm | 10 (2%) | 10 (3%) | 10 (3%) | 6 (60%) | 4 (40%) | ||
| Positive cytology | 109 (17%) | 109 (29%) | 109 (29%) | – | 109 (100%) | ||
| Solid mass | 170 (27%) | 170 (46%) | 170 (46%) | 27 (16%) | 143 (84%) | ||
AI= Absolute indications, RI= Relative indications, HRS= High-risk stigmata, WF= Worrisome features. Patients who had both AI and RI, or HRS and WF, were classified in the category of higher risk when considering IG and EG. MPD= Main pancreatic duct. * 8 of the 627 patients included in the study did not carry any feature of risk and were surgically treated due to pancreatic cancer familiarity or personal preference.
3.2 2018 European guidelines and 2017 International guidelines
Table 3 shows the correlation between final histology (benign and malignant IPMN) and 2018 EG and 2017 IG criteria for surgery. HRS were identical to AI and MPD dilation ≥10 mm and solid mass were the most common features in these two groups. The rate of malignancy was 59.5% in patients with 2018 EG, including both AI and RI, and among malignant IPMNs 66.6% had invasive cancer and 33.4% HGD. The rate of malignant IPMNs was 35.4% for patients with only RI compared with 75.3% for those with AI (p <0.01). PPV of EG considering AI and RI was 59.5% (95%CI: 59–59.9). NPV was calculated on the small population of resected patients without any AI/RI, and it was 62.5% (95% CI: 30.4–86.5). Considering 2017 IG categories, malignancy rates were very close to their 2018 EG counterparts. In HRS positive IPMNs the malignancy rate was 75.3% compared with 35.1% in WF positive IPMNs (p <0.01). PPV and NPV of 2017 IG considering both WF and HRS were 59.3% (95%CI: 58.9–59.7) and 50% (95%CI: 16.9–83.1), respectively.
Table 3–Correlation between final histology (benign and malignant IPMN) and 2018 European Guidelines and 2017 International Guidelines criteria for surgery.
| N of patients | 627 | ||
|---|---|---|---|
| EG +ve (AI+RI) -Benign -Malignant -Invasive -HGD | 619 (98.7%) 251/619 (40.5%) 368/619 (59.5%) 245/368 (66.6%) 123/368 (33.4%) | IG +ve (HRS+WF) -Benign -Malignant -Invasive -HGD | 621 (99%) 253/621 (40.7%) 368/621 (59.2%) 245/368 (66.6%) 123/368 (33.4%) |
| AI -Benign -Malignant | 373/619 (60.3%) 92/373 (24.7%) 281/373 (75.3%) | HRS -Benign -Malignant | 373/621 (60.1%) 92/373 (24.7%) 281/373 (75.3%) |
| RI -Benign -Malignant | 246/619 (39.7%) 159/246 (64.6%) 87/246 (35.4%) | WF -Benign -Malignant | 248/621 (39.9%) 161/248 (64.9%) 87/248 (35.1%) |
| EG –ve -Benign -Malignant | 8 (1%) 5/8 (63%) 3/8 (37%) | IG –ve -Benign -Malignant | 6 (1%) 3/6 (50%) 3/6 (50%) |
| PPV EG | 368/619 (59.5%) CI=59–59.9 | PPV IG | 368/621 (59.3%) CI=58.9–59.7 |
| NPV EG | 5/8 (62.5%) CI=30.4–86.5 | NPV IG | 3/6 (50%) CI=16.9–83.1 |
HGD= High grade dysplasia, AI= Absolute indications, RI= Relative indications, EG= European guidelines, IG= International guidelines, HRS= High risk stigmata, WF= Worrisome features, + ve= Positive, – ve= Negative, PPV= Positive predictive value, NPV= Negative predictive value, CI= 95% confidence interval.
3.3 The role of endoscopic ultrasound in IPMN management
Only 297 (47.4%) patients underwent EUS before surgery (Table 4) as recommended by IG. Overall, 297 (47.4%) patients underwent EUS before surgery, and only 116/297 (39%) had WF that required EUS as recommended by IG. Considering this guideline, EUS confirmed MRI classification of WF/HRS in 260/297 patients (87.5%). Out of 116 patients initially classified as WF-IPMN by MRI, 31 (26.7%) were reclassified as HRS-IPMN after EUS. Of these 31 patients, 24/31 (77.4%) were found to be malignant at final pathology, 10 patients had HGD and 14 had an invasive cancer. Only 5/180 (2.8%) patients with a MRI diagnosis of HRS-IPMN were subsequently downgraded to WF-IPMN by EUS, and 3/5 (60%) had malignancy. There was 1/116 (0.9%) patient with a MRI diagnosis of WF-IPMN that was re-classified as IPMN without WF/HRS by EUS (low-grade dysplasia at final histology).
Table 4The role of endoscopic ultrasound following initial magnetic resonance imaging.
| Patients who underwent EUS | N = 297 | % |
|---|---|---|
| MRI findings | ||
| -HRS | 180/297 | 60.6 |
| -WF | 116/297 | 39.1 |
| -No HRS/WF | 1/297 | 0.3 |
| EUS confirmed MRI findings | 260/297 | 87.5 |
| -EUS confirmed HRS at MRI | 175/180 | 97.2 |
| -EUS confirmed WF at MRI | 84/116 | 72.4 |
| -Malignancy at final histology | 162/260 | 62.3 |
| WF in EUS after negative MRI | 0 | |
| HRS in EUS after WF in MRI | 31/116 | 26.7 |
| - Malignancy at final histology | 24/31 | 77.4 |
| Negative EUS after WF in MRI | 1/116 | 0.9 |
| - Malignancy at final histology | 0 | |
| WF in EUS after HRS in MRI | 5/180 | 2.8 |
| - Malignancy at final histology | 3/5 | 60 |
WF = worrisome features, HRS = High risk stigmata, EUS= endoscopic ultrasound, MRI= magnetic resonance imaging. Malignancy: includes both high grade dysplasia and invasive cancer.
3.4 Comparison of 2017 IG and 2018 EG
Table 5 shows sensitivity, specificity, PPV, NPV and accuracy in predicting malignant IPMNs for both 2017 IG and 2018 EG. There were no differences between HRS and AI, resulting in an accuracy rate of 71%. WFs did not differ significantly when compared to RIs. Both WF and RI had a sensitivity of 96.7% but low specificity (1.8% and 3.1%) for the detection of malignancy, with a low accuracy rate (35.4% vs 36.2%). Similar results were obtained by comparing WF+HRS versus RI+AI (Table 5).
Table 5Sensitivity, specificity, accuracy, positive and negative predictive value of the different categories of risk for malignancy of 2018 European Guidelines and 2017 international Guidelines.
| International Guidelines | European Guidelines | ||||
|---|---|---|---|---|---|
| HRS/AI | WF | HRS+WF | RI | AI+RI | |
| Sensitivity | 75.7% (71.1–80) | 96.7% (90.6–99.3) | 99.2% (97.7–99.8) | 96.7% (90.6–99.3) | 99.2% (97.7–99.8) |
| Specificity | 64.1% (57.9–69.9) | 1.8% (0.4–5.3) | 1.2% (0.2–3.4) | 3.1% (1–7) | 2% (0.6–4.5) |
| PPV | 75.3% (72–78.4) | 35.1% (34.1–36.1) | 59.3% (58.9–59.7) | 35.4% (34.3–36.5) | 59.5% (59–59.9) |
| NPV | 64.6% (59.8–69) | 50% (17.1–82.9) | 50% (16.9–83.1) | 62.5% (29–87.2) | 62.5% (28.7–87.4) |
| Accuracy | 71% (67.3–74.5) | 35.4% (29.6–41.7) | 59.2% (55.2–63.1) | 36.2% (30.3–42.5) | 59.5% (55.5–63.4) |
HRS= High risk stigmata, AI= Absolute indications, WF= Worrisome features, RI= Relative indications, PPV= Positive predictive value, NPV= Negative predictive value. The values in bracket represent 95% confidence intervals.
By strictly applying 2017 IG and 2018 EG to the study population and by considering all 627 patients completely fit for surgery (Table 6), all patients with AI + RI should have undergone immediate surgery according to EG. On the other hand IG would recommend immediate surgery for patients with HRS. Patients with WF should undergo EUS and based on EUS results, either surgery or strict observation could be considered. Therefore, focusing on 2017 IG, we considered in this analysis all patients with HRS plus those with radiological WF that underwent also EUS. In this setting EG resulted in a higher sensitivity (99.2%) compared to IG (83%), but EG also had both a lower PPV (59.5 vs 65.8%) and specificity (2% vs. 37.5%). The overall diagnostic accuracy resulted similar (59.5 vs 64.4%). In the third column of Table 6 we have speculated on IG. We strictly applied the flowchart of IG, considering only a group of 473 patients including all patients with HRS at imaging and those patients with radiologic WF that had a “positive EUS”. In this scenario we considered as appropriate surgical indications HRS-IPMN and WF-IPMN that had an upgrade to HRS-IPMN after EUS (positive EUS). With this retrospective strict application of IG, we obtained an improvement of all diagnostic parameters, including sensitivity (92.9%), specificity (42%), PPV (75.5%), NPV (75.6%) and accuracy (75.5%). In Supplementary Table 1 we translated these results in an “overtreatment” rate of 40% with the use of EG compared to 25.5% of IG (HRS plus WF undergoing EUS). EG were associated with a very low risk (0.5%) of undertreatment, while the risk of missing a malignant IPMN was 10% by applying IG. When we considered only patients with WF who underwent EUS (n = 455 473), the “overtreatment” rate was 19.9% and the under-treatment rate only 4.7%.
= These results were calculated on a subset of 473 patient, simulating a population of patients where 2017 IG flowchart was strictly applied. This subset was extracted excluding from our initial population all the patients without HRS that did not undergo EUS in the pre-operative workup, as 2017 IG recommend. As positive EUS we considered the 31 patients that were classified as WF by MRI and then received an upgrade of surgical indication to HRS after undergoing EUS.
Table 6Comparison of diagnostic performance for malignancy detection between 2017 International Guidelines (IG) and 2018 European Guidelines (EG) by strictly applying the management policy of the two guidelines to a cohort of surgically fit patients.
| N of patients | n = 627 | n = 473 | |
|---|---|---|---|
| 2017 IG: HRS and WF with EUS | 2018 EG: AI and RI | 2017 IG: HRS and WF with positive EUS * = These results were calculated on a subset of 473 patient, simulating a population of patients where 2017 IG flowchart was strictly applied. This subset was extracted excluding from our initial population all the patients without HRS that did not undergo EUS in the pre-operative workup, as 2017 IG recommend. As positive EUS we considered the 31 patients that were classified as WF by MRI and then received an upgrade of surgical indication to HRS after undergoing EUS. | |
| Sensitivity | 83% (78.8–86.7) | 99.2% (97.7–99.8) | 92.9% (89.5–95.5) |
| Specificity | 37.5% (31.6–43.7) | 2% h(0.6–4.5) | 42% (34.3–50) |
| PPV | 65.8% (63.4–68.1) | 59.5% (59–59.9) | 75.5% (72.9–77.9) |
| NPV | 60.4% (53.7–66.7) | 62.5% (28.7–87.4) | 75.6% (66.5–82.8) |
| Accuracy | 64.4% (60.4–68.2) | 59.5% (55.5–63.4) | 75.5% (71.3–79.3) |
HRS= High-risk stigmata, AI= Absolute indications, WF= Worrisome features, RI= Relative indications, PPV= Positive predictive value, NPV= Negative predictive value. The values in bracket represent 95% confidence intervals.
4. Discussion
The increased diagnosis of IPMNs has resulted in the development of several guidelines for their management [
[5]
–[10]
. Among those, IG and EG have been updated based on the most recent evidence from the literature [[12]
,[13]
. Compared with previous (2012) version, IG had only minor revisions made during a consensus meeting among experts [[12]
]. On the other hand, EG underwent a major revision process, based on systematic reviews and GRADE methodology that resulted in “evidence-based guidelines” [[13]
]. In the present paper we aimed to evaluate NPV, PPV, sensitivity, specificity and accuracy of these guidelines to detect patients with malignant IPMNs. We considered a large surgical cohort in order to have a certain histological diagnosis. Both guidelines identify two categories of risk for malignancy, including AI/RI (2018EG) and HRS/WF (2017IG) [[12]
,[13]
. AI and HRS comprise features that are significantly associated with malignancy. Jaundice, MPD >10 mm, enhancing nodules ≥5 mm and cytology positive for high-grade dysplasia/adenocarcinoma are identified as AI and HRS. EG clearly include also a “solid pancreatic mass” within AI [[13]
], and although IG does not explicitly mention it we considered it a straightforward feature of malignancy and thus a HRS. There were no major differences also when comparing WF and RI. The overlapping in the diagnostic criteria of AI/HRS and of WF/RI significantly affected the results of the comparison between the two guidelines. Specifically, AI resulted identical to HRS, with a 75.7% sensitivity but a 64.6% specificity and a 71% accuracy. Both WF and RI had a 96.7% sensitivity but a low specificity (1.8 and 3.1%) in detection of malignancy, with low PPV (35.1 and 35.4%) and accuracy (35.4 and 36.2%).While the two guidelines identify similar diagnostic parameters, the clinical management of patients is different. Both guidelines agree to recommend immediate surgery for patients with HRS/AI, but the management of WF/RI is more controversial [
[12]
,[13]
. In surgically fit patients with no comorbidities, EG suggest resecting IPMNs with a single RI, as EG stress the importance to identify and treat all patients with HGS before development of invasive cancer. IG recommend EUS when WF are evident at radiology, and depending on EUS results patients may undergo surgery or strict surveillance – also considering their age, comorbidities and life expectancy.Therefore, we evaluated (i) the possible impact of EUS on the management of IPMNs patients, and (ii) different management strategies regarding WF/RI. Firstly, 297 patients (43%) underwent EUS after an initial MRI. EUS confirmed MRI findings in 87.5% of cases, including 97.2% of radiological HRS and 72.4% of radiological WF. Interestingly, EUS detected HRS in 31 patients out of 116 (26.7%) who had WF at MRI, thus improving the clinical management in one out of four patients with “radiological” WF, as 24/31 patients (77.4%) had malignancy at final histology. Only 5/180 (2.8%) patients with a radiological diagnosis of HRS-IPMN were classified as WF-IPMN by EUS. Among them, 3/5 (60%) had a malignancy. This result underlines that patients with HRS should undergo immediate surgery without further testing. Nonetheless EUS may improve the diagnostic yield after MRI in patients with WF/RI, although prospective studies are needed to confirm this finding.
Secondly, we tried to apply the different strategies according to EG and IG on patients with WF/RI. By considering a cohort of 627 patients theoretically fit for surgery and without any comorbidity, all patients with AI and RI should have been operated based on 2018 EG [
[13]
]. In the same setting, only patients with HRS plus those with WF who had also EUS would have undergone resection [[12]
]. With this analysis sensitivity, specificity and PPV were 99.2, 2, and 59.5% for EG compared with 83, 37.5 and 64.4% for IG. Therefore, the risk of undertreatment (missing malignant IPMNs) was only 0.5% with EG compared to 10% with IG. On the contrary, the rate of potential overtreatment was 40 and 25.5% by applying EG and IG, respectively. Therefore, by considering a more aggressive policy including surgery for patients with a single RI, it is possible to treat all malignant IPMNs, but four out of ten resected patients will receive a diagnosis of low-grade IPMN and they could have undergone surveillance. Obviously, it is difficult to clearly define an “overtreatment”, because the fate of an IPMN with low-grade dysplasia is uncertain. In a hypothetical scenario of a low-grade IPMN involving the main duct, surgery can be considered appropriate for a patient with a long life-expectancy, but surveillance can be appropriate for an elderly individual as the time required for progression to malignancy exceeds life expectancy. This patient-oriented approach has been strongly underlined in EG. Finally, we strictly applied IG to the cohort, and in our last analysis we did not consider those patients with radiological WF who did not undergo EUS. Thus, we analyzed 473 patients meeting these criteria, and in this specific scenario only patients with radiological HRS and those with WF upgrade to HRS after EUS should have been operated. In this setting sensitivity, specificity, PPV were 92.9%, 42% and 75.5% resulting in an undertreatment of 4.7% and in a possible overtreatment of 19.9%. Despite the limitation of a retrospective analysis, it seems that EUS can have a role in improving the diagnostic yield of IPMNs and to identify those lesions that have an appropriate surgical indication.In very high volume centers, pancreatic surgery-related mortality is usually below 2%, while cancer related mortality is almost equal to its incidence [
[17]
,[18]
. Therefore, every “under-treatment” potentially results in a cancer-related death. If we consider 100 patients operated according to the two guidelines at high-volume centers, the total deaths is considerably different, and it is in favor of the application of the EG. In IG group the mortality rate would be 6.7% (4.7% resulting from undertreatment of IPMNs plus 2% resulting from surgery-related mortality). In EG group the mortality rate would be 2.5% (0.5% resulting from IPMNs under-treatment plus 2% from surgery-related mortality). This theoretical picture would significantly change in low volume centers where mortality following pancreatectomy can be much higher than 2%. Moreover, the potential benefit of “preventive surgery” for low-grade IPMNs should be balanced with the risk of long-term complications (i.e. diabetes), along with the need for surveillance of the remaining pancreas because of methacronous IPMNs or IPMN-independent PDAC [- Bassi C.
- Marchegiani G.
- Giuliani T.
- et al.
Pancreatoduodenectomy at the Verona Pancreas Institute: the evolution of indications, surgical techniques and outcomes. A retrospective analysis of 3000 consecutive cases.
Ann Surg. 2021; (Jan 15Online Ahead of Print)https://doi.org/10.1097/SLA.0000000000004753
[12]
,[13]
,[19]
.Considering the retrospective design of the study with its limitation, it is difficult to make strong recommendations. The role of age, comorbidities and patients’ life expectancy is of paramount importance in the clinical decision-making process, but cannot be considered in retrospective fashion. We considered a surgical cohort in order to have certain pathological data, but we did not consider the outcomes of patients with WF/RI undergoing nonoperative management. This is a limit of our study since the real prevalence of malignancy in a “real life setting” should be evaluated in the entire population with presumed IPMNs, including those under surveillance. Of note, a recent study showed a 5-years disease-specific survival of 96.2% for 231 patients with WF under surveillance supporting the possibility to observe rather than operate all these patients [
[20]
]. In this setting, the diagnostic accuracy of different guidelines should be addressed in prospective cohorts with resected or surveilled patients with IPMNs.It is likely that the tug-of-war between more aggressive and more conservative guidelines will not be solved based on further re-classification of clinical and/or morphological predictors. Molecular testing on cyst fluid or blood will be the key to properly identify IPMNs harboring malignancy or at risk of developing cancer over time [
[21]
,[22]
.Funding
The work was supported by the Lars Vesterlund minnesfond (to J.M.L.) and by Ricerca Finalizzata Ministero della Salute – Giovani Ricercatori “Change-promoting”/GR-2018-12,366,897 (to S.C.)
Declaration of Competing Interest
No author has conflicts of interest to declare.
Appendix. Supplementary materials
References
- Diagnosis and management of pancreatic cystic neoplasms: current evidence and guidelines.Nat Rev Gastroenterol Hepatol. 2019; 16: 676-689
- Intraductal papillary mucinous neoplasms of the pancreas: an updated experience.Ann Surg. 2004; 239: 788-797
- Branch-duct intraductal papillary mucinous neoplasms: observations in 145 patients who underwent resection.Gastroenterology. 2007; 133: 72-799
- Risk of pancreatic malignancy and mortality in branch-duct IPMNs undergoing surveillance: a systematic review and meta-analysis.Dig Liver Dis. 2016; 48: 473-479
- International Association of Pancreatology. International consensus guidelines for management of intraductal papillary mucinous neoplasms and mucinous cystic neoplasms of the pancreas.Pancreatology. 2006; 6: 17-32
- International consensus guidelines 2012 for the management of IPMN and MCN of the pancreas.Pancreatology. 2012; 12: 183-197
- European experts consensus statement on cystic tumors of the pancreas.Dig Liver Dis. 2013; 45: 703-711
- Italian consensus guidelines for the diagnostic work-up and follow-up of cystic pancreatic neoplasms.Dig Liver Dis. 2014; 46: 479-493
- American gastroenterological association institute guideline on the diagnosis and management of asymptomatic neoplastic pancreatic cysts.Gastroenterology. 2015; 148: 819-822
- Quality assessment of the guidelines on cystic neoplasms of the pancreas.Pancreatology. 2015; 15: 463-469
- Effects of comorbidities on outcomes of patients with intraductal papillary mucinous neoplasms.Clin Gastroenterol Hepatol. 2015; 13: 1816-1823
- Revisions of international consensus Fukuoka guidelines for the management of IPMN of the pancreas.Pancreatology. 2017; 17: 738-753
- European Study Group on cystic tumors of the pancreas. European evidence-based guidelines on pancreatic cystic neoplasms.Gut. 2018; 67: 789-804
- Comparison of preoperative conference-based diagnosis with histology of cystic tumors of the pancreas.Ann Surg Oncol. 2014; 21: 1539-1544
Hruban R., Kloppel G., Adler G., et al. Tumors of the pancreas. In: WHO Classification of Tumors of the Digestive System (IARC WHO Classification of Tumors); 2010: 273–279
- A revised classification system and recommendations from the Baltimore Consensus Meeting for neoplastic precursor lesions in the pancreas.Am J Surg Pathol. 2015; 39: 1730-1741
- Evolution of the Whipple procedure at the Massachusetts General Hospital.Surgery. 2012; 152 (Sep): S56-S63
- Pancreatoduodenectomy at the Verona Pancreas Institute: the evolution of indications, surgical techniques and outcomes. A retrospective analysis of 3000 consecutive cases.Ann Surg. 2021; (Jan 15Online Ahead of Print)https://doi.org/10.1097/SLA.0000000000004753
- Predictors of the presence of concomitant invasive ductal carcinoma in intraductal papillary mucinous neoplasm of the pancreas.Ann Surg. 2010; 251: 70-75
- Low progression of intraductal papillary mucinous neoplasms with worrisome features and high-risk stigmata undergoing non-operative management: a mid-term follow-up analysis.Gut. 2017; 66: 495-506
- A multimodality test to guide the management of patients with a pancreatic cyst.Sci Transl Med. 2019; 11
- Preoperative next-generation sequencing of pancreatic cyst fluid is highly accurate in cyst classification and detection of advanced neoplasia.Gut. 2018; 67: 2131-2141
Article info
Publication history
Published online: April 09, 2021
Accepted:
March 4,
2021
Received:
November 4,
2020
Identification
Copyright
© 2021 Published by Elsevier Ltd on behalf of Editrice Gastroenterologica Italiana S.r.l.
User license
Creative Commons Attribution – NonCommercial – NoDerivs (CC BY-NC-ND 4.0) | How you can reuse 
Elsevier's open access license policy
Creative Commons Attribution – NonCommercial – NoDerivs (CC BY-NC-ND 4.0)
Permitted
For non-commercial purposes:
- Read, print & download
- Redistribute or republish the final article
- Text & data mine
- Translate the article (private use only, not for distribution)
- Reuse portions or extracts from the article in other works
Not Permitted
- Sell or re-use for commercial purposes
- Distribute translations or adaptations of the article
Elsevier's open access license policy
ScienceDirect
Access this article on ScienceDirectLinked Article
- Diagnostic accuracy of updated International versus European guidelines for resection of Intraductal papillary mucinous neoplasm: Is it a draw?Digestive and Liver DiseaseVol. 53Issue 8
- PreviewDespite recent updates of the International (IG) (2017) [1] and the European (EG) (2018) [2] guidelines, indications for pancreatic resection of Intraductal papillary mucinous neoplasm (IPMN) are still overestimated. Crippa et al. provides new data supporting this conclusion through their surgical series of 627 patients operated for suspicious of malignant IPMN between 2009 and 2018: 20%−40% of them had actually neither carcinoma nor high-grade dysplasia [3].
- Full-Text
- Preview
