Appraisal of EUS-guided needle-based confocal laser endomicroscopy in the diagnosis of pancreatic lesions: A single Chinese center experience.

BACKGROUND AND OBJECTIVES: In the recent years, EUS is one of the routine procedures in the diagnosis of pancreatic diseases. EUS-guided needle-based confocal laser endomicroscopy (nCLE) is a novel minimally invasive imaging technique in diagnosis of pancreatic diseases. The pilot researches provided us some preliminary findings and conclusions with small samples, low rate of pathological correspondence. The aim of this current study was to evaluate the diagnostic efficacy of EUS-guided nCLE in solid pancreatic lesions (SPLs) and pancreatic cystic lesions (PCLs) based on large samples. The date was obtained on nCLE imaging findings and high rate of correlation with pathology.
MATERIAL AND METHODS: Patients enrolled in the study were underwent EUS-nCLE to achieve the nCLE images and diagnosis. Comparing with the final diagnosis, including surgical histopathological results or cyto-/histopathology through FNA, the efficacy and accuracy of nCLE in diagnosis in solid and cystic pancreatic lesions were evaluated. In other cases, clinical diagnoses were achieved based on the combination with clinical history, image findings and fluid analysis and cytology, by 3 independent committee members strongly agreed with a concordant diagnosis.
RESULTS: Totally 172 patients were enrolled into the study. The overall rate of final diagnosis was about 65% while 50% in cystic lesion. The mean sensitivity, specificity, negative predictive value, positive predictive value and accuracy of the nCLE in diagnosis of PDAC is 90.3%, 89.5%, 93.3%, 85.0% and 90.0% respectively. The efficacy and accuracy of pancreatic cystic lesions were very satisfying and some additional nCLE signs were found, including "black aggregates of cells, forming as gland-like structure, surrounding by fibro and vessels" in neuroendocrine tumors (NETs); "black columnar protrusions near vascular area" in the pseudopapillary solid tumor (SPT); macrophage in tuberculosis (TB) and small aggregate of black regular cells maybe corresponds to ovarian-like stroma in mucinous cystadenoma (MCN). In the study, 20 (11.6%) patients suffered complications, including symptomatic (5.2%) and asymptomatic (6.4%).
CONCLUSIONS: nCLE observation could improve characterization of indeterminate cysts, or confirm the EUS impression, when cytological confirmation is missing. The technique may deliver information to better guide our clinical decisions.

BACKGROUND

In recent years, EUS is one of the routine procedures in the diagnosis of pancreatic diseases. With the fine-needle aspiration (FNA), the cytology acquisition has improved the efficiency of diagnosis significantly in solid pancreatic lesions (SPLs).[1] The sensitivities and specificities of EUS-FNA for the diagnosis of pancreatic cancer were 85%–92% and 96%–98%, respectively.[2] However, EUS-FNA has limitations that include sample error, limited availability of an on-site cytopathologist, additional risk with multiple attempts, and nondiagnostic samples.[3] Furthermore, the accurate diagnosis in pancreatic cystic lesions (PCLs) still poses a challenge for physicians and surgeons, despite applying a combination of advanced imaging technique,[45] FNA cytological examination,[6] and fluid analysis.[7] Hence, the overall management remains suboptimal with unnecessary repeated follow-up or surgery performed.[8]

EUS-guided needle-based confocal laser endomicroscopy (nCLE) is a novel minimally invasive imaging technique, which can be placed into the pancreatic cyst through the 19G fine-needle puncture and provides us the microscopic image of the inner wall of the cyst or parenchyma tissue of the solid tumor, with 1000 times enlargement. Since 2011,[9] EUS-nCLE has been applied in the diagnosis of pancreatic neoplasms and has becoming the focal spot in the field in this decade year. The pilot researches[10111213] of nCLE provided us some preliminary findings and conclusions with small samples, such as superficial vascular network (SVN) in serous cystadenoma (SCA) and finger-like projection in intraductal papillary mucinous neoplasm with low rate of pathological correspondence. The aim of this current study was to evaluate the diagnostic efficacy of EUS-guided nCLE in SPLs and PCLs based on large samples. The date was obtained on nCLE imaging findings and high rate of correlation with pathology. In addition, some new imaging findings were summarized in the study, which may be defined as potential diagnostic criteria.

MATERIALS AND METHODS

Patients

Patients presenting for EUS observation of pancreatic lesions were recruited from Huashan Hospital, Shanghai, China, participating in the study. Patients were provided informed consent for the study. The inclusion criteria were (1) ≥18 years, (2) suspending pancreatic lesions, and (3) plan to perform EUS procedure. The exclusion criteria were (1) patients with pregnant, breastfeeding, or renal insufficiency; (2) allergy to fluorescein; and (3) nCLE image blurring and difficult to recognize. The protocol was reviewed and approved by the institutional review board of the hospital.

Devices and procedure

In the study, EUS (SU-7000 or SU-9000; FUJIFILM, Tokyo, Japan) was used for the EUS observation and following procedure of nCLE. A prototype probe (AQ-Flex 19; Mauna Kea Technologies, Paris, France) was used for nCLE. This nCLE probe has 10,000 optical fibers, a diameter of 0.85 mm, a field of view of 320 μm, a lateral resolution of 3.5 μm, a length of 4 m, and a focal depth of 40–70 mm. Before the procedure, the fibers were profited within a 19-G FNA needle (EchoTip; Cook Medical, Bloomington, Ind) with advancement of the probe tip approximately 2 mm beyond the needle tip. The probe position was then secured using a locking device that attaches the probe at a specific length to the inlet of the biopsy channel.

After the routine examination of EUS, the FNA needle was inserted under EUS guidance. Then, the probe was positioned to extend no more than 2 mm from the tip of the needle, and 2.5 mL of 10% fluorescein was injected simultaneously. The needle was moved using “Fanning Technique” to observation in different positions in solid lesions or different areas of the cyst wall. After nCLE image acquisition, FNA was performed for cytology in the SPLs, while the fluid acquisition was performed in the PCLs for quantitative tumor markers and amylase.

Needle-based confocal laser endomicroscopy diagnosis

nCLE diagnosis was established based on the previous original studies, INSPECT,[11] DETECT,[14] CONTACT Phase I and II,[1315] and others.[16] For example, the criteria for diagnosing malignant SPLs was dark clumps with or without dilated vessels; the criteria for diagnosing SCA was SVN, the finger-like papillary projection, and so on. During the study, we reviewed all the images to correct the nCLE diagnosis and concluded that some additional potential imaging may help in making a diagnosis.

Final diagnosis

The final diagnosis was established by the surgical histological specimen in those patients who underwent resection of the pancreatic lesion. In those patients who did not undergo surgery, the final diagnosis was confirmed by positive FNA cytology. In those patients without positive histo-/cytological results, clinical diagnosis was established by a diagnostic committee, including two endoscopists, two surgeons, and one radiologist, based on combination with clinical history, image findings on EUS, computed tomography (CT), and magnetic resonance imaging, as well as fluid analysis and cytology. Clinical diagnoses were defined as high certainty if three independent committee members strongly agreed with a concordant diagnosis. Otherwise, the diagnosis was defined as uncertain.

Statistical analysis

The Fisher's exact test was used for the assessment of the association between imaging findings and clinical diagnosis. Accuracy (AC), sensitivity (SN), specificity (SP), positive predictive value (PPV), and negative predictive value (NPV) with a 95% confidence interval were calculated. P < 0.05 was considered statistically significant. A Bonferroni adjustment was performed for multiple testing for associations of imaging findings and clinical diagnosis.

RESULTS

A total 172 patients were enrolled in the study from Huashan hospital from October 2016 to October 2018. Table 1 provides general clinical characteristics of the cases. The cases included 50 solid lesions and 122 cystic/mixed cases. Sixty lesions were in the head and uncinate of the pancreas, 102 in the distal pancreas, and 6 cases were multiple lesions in pancreas. The other 4 cases were diffuse lesion in the pancreas. The overall rate of final diagnosis was about 65% (112/172). As to cystic lesions, 50% (61/122) patients were diagnosed by surgical histopathology and 6% (7/122) patients were diagnosed by the FNA cytology.

Table 1

The general characteristics of the patients

Characteristics	Overall (n=172)
Sex, male, n (%)	75 (43.6)
Age, mean y/o	54.2 (19-79)
Average imaging time (mins)	7.92
Location of lesion, n (%)
Head	52 (30.2)
Distal	102 (59.3)
Uncinate	8 (4.7)
Multiple	6 (3.5)
Diffuse	4 (2.3)
Lesion category, n (%)
Cystic	93 (54.1)
Mixed solid and cystic	29 (16.9)
Solid	50 (29.1)
Approach, n (%)
Stomach	119 (69.2)
Duodenal	53 (30.8)
Access of lesion, n (%)
Easy	124 (72.1)
Moderate	39 (22.7)
Difficult	9 (5.2)

Solid lesions

In the study, 88% (44/50) of SPLs were confirmed by the final diagnosis, including 23 surgical histopathologies and 21 FNA cytologies. We had pancreatic ductal adenocarcinoma (PDAC), solid pseudopapillary tumor (SPT), neuroendocrine tumor (NET), tuberculosis (TB), and other rare cases. EUS-nCLE produced satisfactory images for all SPL patients. EUS-nCLE was diagnosed as PDAC in 31 patients, while benign/borderline SPLs in 19 patients. Thirty-one PDAC patients were confirmed based on 16 surgical histopathologies, 14 FNA cytologies, and 1 clinical diagnosis. The 19 benign/ borderline SPLs included 7 surgical histopathological diagnosess, 7 FNA cytological diagnoses, and 5 clinical diagnoses.

In most PDAC cases, the nCLE imaging showed similar findings, which describes as dark clumps of cells larger than 40 μm with or without dilated irregular vessels larger than 20 μm with fluorescein leakage. The dark clumps of cells appeared in the 28 PDAC patients, while the dilated vessels appeared in 22 patients. Both images appeared simultaneously in 19 patients [Figure 1]. The mean SN, SP, NPV, PPV and AC, for the nCLE in the diagnosis of PDAC was 90.3%, 89.5%, 93.3%, 85.0%, and 90.0%, respectively.

Figure 1

Needle-based confocal laser endomicroscopy images of solid pancreatic lesions. (a) Dilated irregular vessels (arrow) larger than 20 μm with fluorescein leakage in pancreatic ductal adenocarcinoma; (b) Dark clumps of cells (arrow) in pancreatic ductal adenocarcinoma; (c) Black aggregates of cells (arrow), forming as gland-like structure; (d) Black columnar protrusions (arrow) may be correspond to pseudopapilla in solid pseudopapillary tumor; (e) Massive fibro area in chronic pancreatitis; (f) Huge black cell (maybe macrophage) in tuberculosis

Within the other 19 cases, we had 14 definite diagnosess and 5 clinical diagnosies, including 5 NETs, 3 SPTs, 3 accessory spleens, 3 autoimmune pancreatitis (AIP), 1 chronic pancreatitis (CP), 2 lymphomas, 1 gastrointestinal stromal tumor, and 1 TB.

In the NET cases, we observed black aggregates of cells, forming as gland-like structure, surrounding by fibrovessels. In SPT cases, we found black columnar protrusions near the vascular area. In AIPs and CP cases, massive fibrous areas were observed with scattered cells or glands. In the TB case, black huge cells, may corresponds to macrophage, were found easily, mixing with vesicular adipocytes [Figure 1].

Cystic lesions

Totally, we have 122 cases of PCLs and mixed type of lesions. The categories and diagnostic methods of cases in the study were listed in Table 2.

Table 2

Overall and diagnosis distribution of cystic lesions

Characteristics	n=122
Final diagnosis achieved, n
Surgery	61 (50.0%)
FNA	8 (6.6%)
Clinical agreement	50 (41.0%)
Inclusive	3 (2.4%)
	Total	Operation	FNA	Surveillance
Diagnosis, n
SCN	38	14	/	24
MCN	18	15	/	3
IPMN	31	14	/	17
IPMN with malignancy	3	2	1	1
Pseudocyst	9	3	1	6
SPT	7	6	1	1
PDAC	4	2	3	NA
NET	3	3	1	0
Tuberculosis	1	/	1	1
Other	3	2	/	1
Inclusion	5	/	/	5

FNA: Fine-needle aspiration; SCN: Serous cystic neoplasm; MCN: Mucinous cystic neoplasm; IPMN: Intraductal papillary mucinous neoplasm; SPT: Pseudopapillary solid tumor; PDAC: Pancreatic ductal adenocarcinoma; NET: Neuroendocrine tumors.

We had 38 SCAs based on the final and clinical diagnosis. The typical nCLE images of SCA are called fern-like SVN, which was observed in 81.6% (31/38) of the cases [Figure 2]. Surgical resection was performed in 36.8% (14/38) of the SCA patients. According to the pathological results, nCLE achieved the correct diagnosis in 10 cases, while nCLE was misdiagnosed in two cases and inclusive in two cases. The diagnostic efficacy and AC of nCLE in SCA is listed in Table 3.

Figure 2

Needle-based confocal laser endomicroscopy images of pancreatic cystic lesions. (a) Superficial vascular network in serous cystadenoma; (b) Finger-like projection in intraductal papillary mucinous neoplasm, two dark, parallel epithelial borders (yellow arrow) with a bright vascular core (white arrow); (c) Lots of dark and bright particles in pseudocyst; (d) Single epithelial borders in mucinous cystadenomas; (e) Dilated vessels in mucinous cystadenomas; (f) Small aggregate of black regular cells maybe corresponds to ovarian-like stroma in mucinous cystadenomas

Table 3

Efficacy of nCLE in diagnosis of different PCLs

Type	n	SN (%)	SP (%)	PPV (%)	NPV (%)	AC (%)
SCA	38	89.5	100	100	95.5	96.7
MCN	15	86.7	98.1	86.7	98.1	94.3
IPMN	31	96.7	100	100	98.9	99.2
Overall	122	94.3	98.1	100	94.3	96.7

SN: Sensitivity; SP: Specificity, PPN: Positive predictive value; NPV: Negative predictive value; AC: Accuracy; MCN: Mucinous cystic neoplasm; IPMN: Intraductal papillary mucinous neoplasm; nCLE: needle-based confocal laser endomicroscopy; PCLs: Pancreatic cystic lesions.

Thirty one cases were diagnosed with IPMN. The typical nCLE images of IPMN were described as finger-like structure, which was micropapillary projection composed of two dark, parallel epithelial borders with a bright vascular core [Figure 2]. Among them, 45.2% (14/31) of the cases went to surgical resection. Comparing with the histopathological diagnosis, 92.9% (13/14) of the cases reached a consensus based on the nCLE diagnosis. We also had 18 cases of mucinous cystadenomas (MCA), and 83.3% (15/18) cases were performed the surgical resection. We had 1 misdiagnosis and 1 inclusive case. The diagnostic criteria of nCLE in MCA was described as dilated vessels (φ>15 μm) and/or a single black epithelial border [Figure 2].

We had 9 cases of pseudocysts in the study, only 33.3% cases had pathological diagnosis, while the others were diagnosis based on the clinical information. The nCLE images showed lots of dark and bright particles suspending in cystic fluid, meanwhile without vascular structure or cells on the wall [Figure 2].

As to other mixed cases, the nCLE image characteristics of the solid parts helped in making a diagnosis, which is similar with the SPLs such as SPT, NET, and PDAC.

Complication

In the study, 20 (11.6%) patients suffered complications, including symptomatic and asymptomatic. Pancreatitis (n = 6, 4 mild + 1 moderate + 1 severe) occurred with symptoms; 2 fluid leakage or collection were detected by CT scan; and blooding result in hematoma appeared in 1 patient. Asymptomatic complications were usually found during surgery including intracystic bleeding (n = 4), subserosal hematoma (n = 3), fluid collection (n = 2), and peripancreatic necrosis (n = 1).

DISCUSSION

Pancreatic lesions vary in the clinic, from the lethal malignant tumor to asymptomatic serous cyst. The appropriate treatment depends on the accurate diagnosis. With the advanced imaging techniques, most pancreatic lesions can be recognized. With the EUS-FNA biopsy, the diagnostic AC of SPLs has increased significantly from 62% to 96%,[1] but it still leaves 8%–25% uncertain pathological diagnosis.[17] On the other hand, the accurate diagnosis of PCLs poses a tough challenge to physicians and surgeons. Cross-sectional imaging, EUS, and fluid analysis are currently performed to make a diagnosis. However, the diagnostic AC of PCLs is far from satisfying.[18]

Since 2010, EUS-nCLE has been used to observe the pancreas in models and been proved to be feasible to diagnose pancreatic diseases.[19] Giovannini et al.[10] first described the nCLE, SPL, and PDAC, which show the leakage of fluorescein from irregular vessels in the tumor and large dark clumps representing groups of malignant cells in nCLE images. Afterward, Kongkam et al.[16] indicated the similar signs of malignant SPLs while “find white fibrous bands and normal acinar cells” in benign SPLs. In our study, the similar signs were observed in all the PDAC patients in nCLE images, which diagnostic AC is about 90%. However, EUS-FNA with cyto-/histopathological examination was proved to be satisfying in the diagnosis in SPLs. nCLE did not show certain advantage to FNA in SPLs diagnosis, which can be a supplement when FNA achieves negative result in some inclusive cases. Furthermore, the descriptions of other SPLs, such as NET, were reported[20] based on small samples. We also found some signs that may help in the diagnosis of NET, SPT, TB, and pancreatitis, but more convincing conclusion should be achieved in further studies.

When we focused on the performance of nCLE in PCLs, many surprising findings were satisfying. The most challenging problem in the diagnosis of PCLs is to differentiating between the serous and mucinous cysts. In many studies,[1114212223] nCLE provides AC from 71% to 95%. Compared to these trials, our study consisted of nearly half (43/87, 49.4%) of cystic (SCA, MCA, and IPMN) patients with confirmed histopathology. Description of SCA in nCLE, fern-like SVN, has perfect SP and PPV. In the absence of this pattern, the SN for the diagnosis of SCAs remains low.[1215] Since the epithelium of oligocystic SCAs can become denuded, they can be difficult to distinguish when the characteristic vascular pattern is not observed.[24] Konda et al.[911] described the nCLE criteria of IPMN and mucinous cystic neoplasm (MCN) and analyzed the different images between IPMN subtypes. The finger-like projection has perfect PPV in the diagnosis of IPMN, but similar to SCA, the SN is not high. In these studies, MCN has relatively vague criteria, a gray band with a thin dark line, representing the epithelial border of the cyst wall, with or without deep blood vessels. Sometimes, it is difficult to differential with IPMN in nCLE, especially with untypical IPMN. The combination of nCLE imaging with EUS-FNA and cyst fluid tests may contribute to the differential diagnosis of mucinous cysts.[18] Compared to previous studies, our study enrolled more samples with a higher rate of the final diagnosis. The diagnostic efficacy and AC of nCLE may be more convincing [Table 4]. Mopreover, we have some additional findings not mentioned in previous studies, such as small aggregate of black regular cells may be correspond to ovarian-like stroma, which is one of the defining histological features in MCA [Figure 2].

Table 4

Efficacy of nCLE in diagnosis in PCLs in previous studies

Study	Year	Outcome	Patients, n	Surgery n (%)	SN (%)	SP (%)	PPV (%)	NPV (%)	Accuracy (%)
INSPECT	2013	PCLs	66	14 (21.2)	59	100	100	50	71
DETECT	2015	Mucinous cystic lesion	30	2 (6.0)	80	100	100	80	89
CONTACT-1	2015	SCA	31	7 (22.5)	69	100	100	82	87
CONTACT-2	2015	Mucinous cystic lesion	33	9 (27.3)	91	95	91	95	94
INDEX	2016	Mucinous cystic lesion	26	20 (76.9)	94	82	89	93	89
Napoléon et al.	2018	PCLs	79	38 (50.0)	95	100	100	96	98
Our Study	2018	PCLs	122	61 (50.0)	94	98	100	94	97

SN: Sensitivity; SP: Specificity; PPN: Positive predictive value; NPV: Negative predictive value; nCLE: needle-based confocal laser endomicroscopy; PCLs: Pancreatic cystic lesions.

The last thing we have to mention is about the complications of nCLE, previous papers proved that the puncture in cystic lesion will cause more complications than in solid tumors, and the miniprobe of confocal laser could only pass through 19G needle, not the common 22 and 25G. If we define the complication is symptomatic events, the rate is 5.2%, which is similar with previous studies. Furthermore, there still have some complications without any obvious symptoms, we would like to name them potential complications, not delay the following treatment, and only could be found during the operation. Hence, we must notice that the incidence of complications in EUS-nCLE is higher than that in EUS-FNA.

CONCLUSION

The utilization of nCLE in pancreatic cysts is still at the very beginning period; some of our patients really benefit from this technique. Confocal observation could improve the characterization of indeterminate cysts or confirm the EUS impression, when cytological confirmation is missing. The technique may deliver information to better guide our clinical decisions.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.