Advice

Appendix

Appendix

Contents

Data tables

Table 1: Overview of the Konda et al. (2011) study

Table 2: Overview of the Konda et al. (2013) study

Table 3: Overview of the Nakai et al. (2015) study

Table 4: Overview of the Napoleon et al. (2015a) study

Table 5: Overview of the Napoleon et al. (2015b) study

Table 1 Overview of the Konda et al. (2011) study

Study component

Description

Objectives/hypotheses

To evaluate the feasibility of nCLE during EUS‑FNA of pancreatic lesions.

Study design

Single-centre, prospective, single-arm feasibility and safety study.

Setting

One tertiary care centre in the US. Dates of data collection not given. No follow‑up period was specified.

Inclusion/exclusion criteria

Inclusion:

  • people having EUS‑FNA of pancreatic lesions.

Exclusion:

  • people with a previous allergic reaction to fluorescein

  • people with renal insufficiency

  • women who were pregnant or breastfeeding.

Primary outcomes

Feasibility and safety.

Statistical methods

Descriptive statistics were used to describe technical feasibility and safety.

A 5‑point Likert scale was used to evaluate image quality.

Patients included

n=18 (7 men, 11 women; mean age 57.9 years).

Results

Images were obtained in 17 of 18 people. There were no device malfunctions. Technical challenges were encountered when imaging 6 of the 18 patients.

Image quality was rated as good to very good in 10 of the 17 patients. Images from 5 patients were rated as moderate quality. Images were rated as low or poor quality for 2 patients.

Sixteen of 18 patients had pancreatic cysts.

Two serious adverse events occurred; both were pancreatitis needing hospitalisation.

Conclusions

nCLE of the pancreas is technically feasible via a 19‑gauge needle under endosonographic guidance.

Abbreviations: nCLE, needle-based confocal laser endomicroscopy; EUS‑FNA, endoscopic ultrasound-guided fine needle aspiration.

Table 2 Overview of the Konda et al. (2013) study

Study component

Description

Objectives/hypotheses

To assess both the diagnostic potential of nCLE in differentiating types of pancreatic cysts and the safety of the technique.

Study design

Single-arm, prospective, multicentre pilot study.

Setting

Patients were enrolled from 8 referral centres (6 in the US and 2 in Europe), between July 2010 and September 2011. No follow‑up period was specified.

Inclusion/exclusion criteria

Inclusion:

  • people having EUS‑FNA for a pancreatic cyst and who were being considered for surgery.

Exclusion:

  • people with renal insufficiency

  • people with an allergy to fluorescein.

Primary outcomes

Diagnostic accuracy, sensitivity, specificity, PPV, NPV. Safety. Clinical diagnosis was made by clinical consensus (n=43) or histopathology (n=14).

Statistical methods

Fisher's exact test was used to examine the association between pancreatic cysts diagnosed by surgical and non-surgical methods and features found by nCLE.

Exact 95% CIs were constructed based on the binomial distribution.

Descriptive statistics were carried out on safety data.

Patients included

n=57 (32 men, 25 women; mean age 63.1 years).

Results

Results from imaging in 26 patients were used to develop diagnostic terms for structures identified on nCLE.

In 31 remaining patients, the presence of epithelial villous structures seen using nCLE was associated with potentially malignant cysts (p=0.004) and provided a diagnostic accuracy of 71% (95% CI 52 to 86), with a sensitivity of 59% (95% CI 36 to 79), specificity of 100% (95% CI 66 to 100), PPV of 100% (95% CI 75 to 100) and NPV of 50% (95% CI 26 to 74). No other features were significantly associated with potentially malignant cysts.

The overall complication rate was 9% and included pancreatitis (n=1 mild, n=1 moderate), transient abdominal pain (n=1), and intracystic bleeding not requiring any further measures (n=3).

Conclusions

The authors concluded that the preliminary data suggest that nCLE has a high specificity in the detection of potentially malignant cysts, but may be limited by low sensitivity.

Abbreviations: CI, confidence interval; EUS‑FNA, endoscopic ultrasound-guided fine needle aspiration; nCLE, needle-based confocal laser endomicroscopy; NPV, negative predictive value; PPV, positive predictive value.

Table 3 Overview of the Nakai et al. (2015) study

Study component

Description

Objectives/hypotheses

To assess the feasibility, safety and diagnostic accuracy of combined nCLE and cystoscopy in the clinical diagnosis of potentially malignant cysts.

Study design

A single-centre, prospective, single-arm feasibility study.

Setting

An academic tertiary referral centre in the US. Dates of data collection not given. There was a follow‑up period of 6 months to 2 years.

Inclusion/exclusion criteria

Inclusion criteria:

  • people over 18 years with a pancreatic cyst.

Exclusion criteria:

  • people contraindicated for EUS‑FNA such as patients who were medically unstable, severe coagulopathy or poor visualisation on EUS for various reasons

  • women who were pregnant or breast-feeding

  • allergy to fluorescein

  • people with renal insufficiency.

Primary outcomes

Technical feasibility and safety. Associations of cystoscopy and nCLE findings with clinical diagnosis of potentially malignant cysts. Clinical diagnosis was made by clinical consensus (n=28) or histopathology (n=2).

Statistical methods

Fisher's exact test was used to assess the association between imaging findings and clinical diagnosis. Diagnostic accuracy, sensitivity, specificity, PPV and NPV with a 95% CI were calculated. A Bonferroni adjustment was performed for multiple testing for associations of imaging findings and clinical diagnosis.

Patients included

n=30 (9 men, 21 women; mean age 72 years [range 37 to 86 years]).

Results

The procedure was technically successful with the exception of 1 probe exchange failure. Image acquisition was 100% successful, with 90% of images being of excellent quality for nCLE and 67% for cystoscopy.

Pancreatitis developed in 2 patients (7%) after the procedure.

Specific features associated with the clinical diagnosis of mucinous cysts were identified: mucin on cystoscopy and papillary projections and dark rings on nCLE.

Accuracy: All patients

nCLE (%)

Cystoscopy (%)

Diagnostic accuracy

87 (range 72 to 87)

83 (range 69 to 83)

Sensitivity

77 (range 64 to 77)

71 (range 58 to 71)

Specificity

100 (83 to 100)

100 (83 to 100)

PPV

100 (83 to 100)

100 (82 to 100)

NPV

77 (64 to 77)

72 (60 to 72)

Accuracy: Subgroup of high certainty patients for mucinous cysts (n=18)

nCLE (%)

Cystoscopy (%)

Diagnostic accuracy

89 (range 68 to 89)

94 (range 75 to 94)

Sensitivity

80 (range 62 to 80)

90 (range 72 to 90)

Specificity

100 (77 to 100)

100 (78 to 100)

PPV

100 (77 to 100)

100 (80 to 100)

NPV

80 (62 to 80)

89 (69 to 89)

Conclusions

The imaging of pancreatic cysts by nCLE is useful in the clinical diagnosis of potentially malignant cysts.

Abbreviations: CI, confidence interval; EUS‑FNA, endoscopic ultrasound-guided fine needle aspiration; nCLE, needle-based confocal laser endomicroscopy; NPV, negative predictive value; PPV, positive predictive value.

Table 4 Overview of the Napoleon et al. (2015a) study

Study component

Description

Objectives/hypotheses

To identify and validate new diagnostic criteria for nCLE of pancreatic cysts and assess the diagnostic accuracy of nCLE to detect benign cysts (SCA).

Study design

Prospective, 3‑centre, single-arm pilot study.

Setting

Three hospitals in France. Screening was carried out between June 2012 and March 2013. There was a follow‑up period of 1 year.

Inclusion/exclusion criteria

Inclusion criteria:

  • a single cyst with diameter >20 mm and with at least 1 cavity ≥13 mm in size.

Exclusion criteria:

  • adults over 18 years

  • people with a known allergy to fluorescein

  • pregnancy

  • people having an EUS‑FNA procedure within the past 3 months

  • people with chronic calcifying pancreatitis

  • people presenting with criteria for malignancy (tissue mass, metastases, ascites, vascular infiltration).

Primary outcomes

Descriptive criteria for pancreatic cysts; diagnostic accuracy of nCLE for benign cysts (SCA). Clinical diagnosis was made by clinical consensus (n=13), positive FNA (n=11) or histopathology (n=7).

Statistical methods

Diagnostic accuracy, sensitivity, specificity, PPV and NPV with a 95% CI were calculated.

The interobserver variability was calculated using Fleiss' kappa statistics.

Patients included

n=31 (6 men and 25 women; mean age 54 years [range 24 to 76 years]).

Results

A 'superficial vascular network' was identified as an nCLE criterion associated with benign cysts.

Overall diagnostic accuracy (%; n=31):

  • diagnostic accuracy 87 (range 69 to 100)

  • sensitivity 69 (range 44 to 92)

  • specificity 100 (range 100 to 100)

  • PPV 100 (range 100 to 100)

  • NPV 82 (range 66 to 98).

Subgroup analysis of patients with high certainty final diagnosis (%; n=18)a:

  • diagnostic accuracy 94

  • sensitivity 86

  • specificity 100

  • PPV 100

  • NPV 92.

Subgroup analysis for patients with lower certainty final diagnosis (%; n=13)a:

  • diagnostic accuracy 77

  • sensitivity 50

  • specificity 100

  • PPV 100

  • NPV 70.

Images were successfully acquired in all 31 people, with no technical failures reported. Mild acute pancreatitis was reported in 1 person.

The interobserver agreement was substantial, kappa=0.77.

Conclusions

The authors conclude that the new nCLE criterion appears highly specific for the diagnosis of benign pancreatic cysts.

Abbreviations: CI, confidence interval; FNA, fine needle aspiration; nCLE, needle-based confocal laser endomicroscopy; NPV, negative predictive value; PPV, positive predictive value; SCA, serous cystadenoma.

a Ranges not given.

Table 5 Overview of the Napoleon et al. (2015b) study

Study component

Description

Objectives/hypotheses

To determine new nCLE criteria for the diagnosis of pancreatic cysts, to propose a comprehensive nCLE classification for their characterisation, and to validate the performance of these criteria.

Study design

Retrospective analysis of single-arm prospective pilot study.

Setting

Three hospitals in France. Screening was carried out between June 2012 and March 2013. There was a follow‑up period of 1 year.

Inclusion/exclusion criteria

Inclusion criteria:

  • a single cyst with diameter >20 mm and with at least 1 cavity ≥13 mm in size.

Exclusion criteria:

  • adults over 18 years

  • people with a known allergy to fluorescein

  • pregnancy

  • people having an EUS‑FNA procedure within the past 3 months

  • people with chronic calcifying pancreatitis

  • people presenting with criteria for malignancy (tissue mass, metastases, ascites, vascular infiltration).

Primary outcomes

Diagnostic yield and interobserver agreement. Clinical diagnosis was made by clinical consensus (n=13) or histopathology (n=20).

Statistical methods

Diagnostic accuracy, sensitivity, specificity, PPV and NPV with a 95% CI were calculated.

The interobserver variability was calculated using Fleiss' kappa statistics.

Patients included

n=33. Thirty-one of these people were also included in the Napoleon et al. (2015a) study, see table 4. No further demographic information was provided for the additional 2 patients.

Results

New nCLE criteria were identified for MCN, PC and cystic NEN.

nCLE allowed a conclusive result to be drawn in 74% of patients. Of these, 87% were correct and 13% were incorrect when compared with the final diagnosis.

Interobserver agreement for cyst types was (kappa):

  • MCN: 0.35

  • IPMN: 0.56

  • PC: 1.00

  • SCA: 0.88

  • all cysts: 0.72.

Conclusions

The authors concluded that the new interpretation criteria with nCLE could facilitate the diagnosis of pancreatic cyst types.

Abbreviations: CI, confidence interval; NEN, neuroendocrine neoplasm; IPMN, intraductal papillary mucinous neoplasm; MCN, mucinous cystic neoplasm; nCLE, needle-based confocal laser endomicroscopy; PC, pseudocyst; SCA, serous cystadenoma.