Role of tomoelastography in differentiating between non-functional pancreatic neuroendocrine neoplasms and solid pseudopapillary neoplasms

doi:10.21203/rs.3.rs-4266676/v1

Download PDF

Research Article

Role of tomoelastography in differentiating between non-functional pancreatic neuroendocrine neoplasms and solid pseudopapillary neoplasms

https://doi.org/10.21203/rs.3.rs-4266676/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Background

The mechanical properties of pancreatic diseases provide new insight for diagnosis and differentiation. Tomoelastography, a multifrequency MR elastography technique, provides high-resolution maps of shear wave speed (c in m/s) and phase angle (φ in rad) for evaluating pancreatic stiffness and fluidity. We explore the diagnostic performances of stiffness and fluidity quantified using tomoelastography for differentiating between non-functional pancreatic neuroendocrine neoplasms (NF-pNENs) and solid pseudopapillary neoplasms (SPNs).

Methods

This prospective study consecutive enrolled 92 untreated patients with pathologically confirmed NF-pNENs and SPNs who underwent tomoelastography examinations from September 2021 to September 2023. Two radiologists independently measured the stiffness and fluidity, and intra/interobserver agreements were assessed. Multivariate logistic regression analysis was performed to identify independent relevant clinical and radiological features to differentiate NF-pNENs and SPNs. The diagnostic performances of stiffness and fluidity and clinical features for tumor differentiation were evaluated using receiver operating characteristic curves.

Results

Thirty patients with SPNs; 62 with NF-pNENs. The radiologists showed substantial or near-perfect interobserver agreement in evaluating clinical and radiological features. SPNs had lower stiffness (1.87 vs 2.40 m/s, P < 0.001) and fluidity (0.97 vs 1.10 rad, P < 0.001) than NF-pNENs. In the multivariate analysis, the independent relevant factors for tumor differentiation were age (P = 0.002) and stiffness (P < 0.001). The areas under the curves (AUCs) of age and stiffness for tumor differentiation were 0.780 (cutoff, 47.5 years) and 0.876 (cutoff, 2.07 m/s), respectively. The differentiation performance of the combined model (c + age) was better than that of convention model (age + enhancement pattern) (AUC = 0.921 vs 0.813; P = 0.009), as well as compared to the age (AUC = 0.921 vs 0.780; P < 0.001), enhancement pattern (AUC = 0.921 vs 0.675; P < 0.001), and fluidity (φ) (AUC = 0.921 vs 0.788; P = 0.011) metrics, but the stiffness (c) metric alone had comparable differentiation performance (AUC = 0.921 vs 0.876; P = 0.115).

Conclusions

Tomoelastography quantified lesion stiffness values combined with clinical age metrics were effective in identifying NF-pNENs and SPNs, establishing the value of tomoelastography in the non-invasive preoperative quantitative identification of pancreas-associated neoplasms.

pancreatic neoplasm

neuroendocrine tumor

magnetic resonance imaging (MRI)

elastography

Pancreatic neuroendocrine neoplasms (pNENs) are a class of malignant heterogeneous pancreatic tumors originating from pancreatic neuroendocrine cells, with diverse manifestations, including solid, cystic or mixed (solid-cystic) lesions [1, 2]. pNENs can be classified into functional and non-functional based on whether they have corresponding hormone related symptoms or syndromes. Solid pseudopapillary neoplasms (SPNs) are tumors with low malignancy potential characterized by solid and cystic components inside a fibrous pseudocapsule and are seen primarily in young female patients [3]. Non-functional pNENs (NF-pNENs) and SPNs lack specific clinical features and share similar imaging manifestations in the early stage, making them difficult to differentiate using conventional imaging [4]. However, the treatment strategies of NF-pNENs and SPNs differ greatly. Generally, surgery is the preferred and optimal treatment for SPNs, and local pancreatectomy is commonly used to maximize protecting pancreatic function [5]. However, for NF-pNENs less than 2 cm with no metastasis or regional lymph node invasion, immediate surgery may be unnecessary, but regular follow-up is required [6]. For NF-pNENs large than 2 cm (grades 1 and 2), a standard pancreatectomy (pancreaticoduodenectomy or distal pancreatectomy) with regional lymphadenectomy, which has a larger resection scope than that of SPNs, is recommended [7]. Thus, clinicians must distinguish between NF-pNENs and SPNs to avoid administering inappropriate treatment.

Endoscopic ultrasound-guided fine needle aspiration enables obtaining a pathological diagnosis before treatment, but is limited in the dependence on the operator’s experience level, differences in sampling sites and invasive characteristics [8]. Although non-invasive biological markers are helpful for diagnosis, there are still shortcomings. Plasma neuron-specific enolase and chromogranin A are elevated in some patients with NF-pNENs, but the clinical utility of chromogranin A is limited by many factors [9, 10]. Studies have demonstrated that neuron-specific enolase has only 39–43% sensitivity and 65–73% specificity for diagnosing neuroendocrine tumors [11, 12]. Consequently, diagnostic biometrics for neuron-specific enolase are poor. The overlapping imaging features between NF-pNENs and SPNs make it is challenging to differentiate between them using only CT or MRI [4, 13]. Imaging by 68Ga/64Cu-DOTA-somatostatin analog PET-CT can help diagnose NF-pNENs, but their some DOTA-negative NF-pNENs and substantially increases the radiation dose [5]. Therefore, new noninvasive diagnostic methods are needed to differentiate the two tumors.

The mechanical properties of pancreatic diseases provide new insight for diagnosis and differentiation. SPNs have large areas composed of pseudopapillary structures, indicating cells with cellular dyscohesive degeneration [14]. Characteristic myxoid clear material surrounds the delicate papillae, aggregated with foamy histiocytes [15]. NF-pNENs often show variable degrees of fibrosis [16], and some show extensive interstitial fibrosis with high levels of collagen extracellular matrix [17]. Changes in cell adhesion and fibrosis alter the tissue's mechanical properties, providing a biological basis for identifying both tumors. Tomoelastography is a multifrequency MR elastography (MRE) technique with noise-robust data postprocessing that provides high-resolution parameter maps and quantifies tissue stiffness and fluidity [18]. Quantifying stiffness allows evaluating fibrosis in tissues [19], and fluidity reflects the viscous behavior of biological tissues [20]. To the best of our knowledge, the use of tomoelastography in characterizing SPNs remains unexplored.

Therefore, we prospectively investigated the stiffness and fluidity of NF-pNENs and SPNs using tomoelastography and determined the diagnostic performance of tomoelastography in differentiating between NF-pNENs and SPNs.

Study population

The institutional review board of our hospital approved this single-center, prospective study. All patients provided written informed consent. Patients with pathologically confirmed NF-pNENs and SPNs who underwent pancreatic tomoelastography examinations within 14 days before surgery between September 2021 and September 2023 were consecutive enrolled. According to the World Health Organization 2019 classification criteria, NF-pNENs were graded and staged into four categories: neuroendocrine tumor-grade 1, neuroendocrine tumor-grade 2, neuroendocrine tumor-grade 3, and neuroendocrine carcinoma.

Exclusion criteria were 1) patients with poor imaging quality due to respiratory motion artifacts (n = 4); 2) patients who received preoperative puncture biopsies or other treatments such as chemotherapy before examination (n = 4); and 3) patients with incomplete MRE data (n = 3). The final study cohort included 92 patients (Fig. 1).

MRI and multifrequency MRE

Multifrequency MRE was performed in a 3-Tesla MRI scanner (MAGNETOM Prisma; Siemens Healthcare, Erlangen, Germany). All patients were examined in the supine position after fasting for 6–8 hours. The protocol was identical for all patients and included coronal half-Fourier acquisition single-shot turbo spin-echo breath-hold T2-weighted imaging (T2WI), axial trigger turbo spin-echo fat-saturated (fs) T2WI, axial multi-echo Dixon volume-interpolated breath-hold, dynamic MR images and pancreatic tomoelastography. The arterial phase was determined using the real-time MRI fluoroscopic monitoring technique and was obtained 8 seconds after the contrast material arrived in the distal thoracic aorta. Portal venous and delayed-phase were acquired approximately 60 s and 2 min after starting contrast material administration [13, 21]. Table 1 details the protocol parameters, excluding the tomoelastography setup.

Before administering the intravenous contrast agent, four drive frequencies of 30, 40, 50 and 60 Hz were induced via four small plastic driver pads driven by compressed-air pulses with a 0.6-bar amplitude (8.0 cm × 4.0 cm × 1.0 cm) placed on the surface projection of the pancreas (2 pads on the anterior side and 2 on the posterior side) [22, 23]. Multifrequency wave field data were acquired using a single-shot, spin-echo, echo planar imaging sequence with flow-compensated motion-encoding gradients during free-breathing, with frequencies of 37.20, 37.20, 37.48 and 44.88 Hz, using a motion-encoding gradient amplitude of 45 mT/m. The entire pancreas was scanned in 7 minutes and 22 seconds, covered by 35 contiguous axial sections.

Table 1

Sequence parameters of the pancreatic MRI protocol
Sequence	T2-HASTE-bh	T2-trigger-tse-fs	VIBE-q-Dixon-bh	DWI	Multiphase-CE
Plane	Coronal	Axial	Axial	Axial	Axial
TR/TE (ms)	1200 /78	2000/77	9.00/1.05–7.38	3700/47	2.75/1.05
FA (°)	160	103	4	-	12.5
Section Thickness (mm)	4	4	2	5	2
FOV (mm)	380	380	426	380	380
Acquisition Time (s)	31	180–300	20	76	8/35
Acquisition Matrix	168×256	288×384	111×160	100×134	192×320
Acceleration Factor	3	3	4	2	4
Receiver Bandwidth (Hz/Px)	1221	766	1080	2488	710

TR, repetition time; TE, echo time; FA, flip angle; FOV, field of view; bh, breath-hold; tse, turbo spin-echo; fs, fat-saturated; HASTE, half-Fourier acquisition single-shot turbo spin-echo; VIBE, volume-interpolated breath-hold; DWI, diffusion-weighted imaging; CE, contrast enhanced.

Tomoelastography postprocessing

The MRE data were processed with the tomoelastography pipeline, publicly available at https://bioqic-apps.com [23]. Wave number-based multifrequency with in-plane wave numbers and multifrequency dual elasto-visco inversion was used for image reconstruction, producing high-resolution elastograms from multifrequency and multi-component shear wave data [18]. This algorithm generated full field-of-view high-spatial resolution maps of shear wave speed (c) and loss angle of the complex shear modulus (φ). Shear wave speed (c) is referred to as a surrogate marker of stiffness because it is directly linked to the magnitude of the complex shear modulus [24]. Loss angle of the complex shear modulus (φ) refers to the fluidity properties of the tissue, ranging from 0 (pure solid) to π/2 (pure fluid) [22]. Here, we use c and φ to report quantitative measures, and “stiffness” and “fluidity” describe the mechanical properties of the pancreas.

Image analysis

For the image analysis, two radiologists with 7 years (reader 1) and 13 years (reader 2) of experience in pancreatic diagnostic imaging blindly reviewed the images and manually drew the region of interest (ROI) in stiffness maps. The ROIs were at least 100 mm². Three ROIs were placed for each tumor along the tumor margin including the superior margin, at the level of its maximum axial diameter and its inferior margin, which avoided calcification, blood vessels, necrotic areas and visible pancreatic duct dilation. The mean values of each ROI were recorded to obtain overall measurements for each lesion. The tumor boundaries were more visible on standard T1- and T2-weighted MRI than on stiffness maps. Therefore, T1WI and T2WI were used for visual guidance to identify the approximate location of the lesion [23]. Readers 1 and 2, who were blinded to the histopathological and clinical results, measured the pancreatic stiffness and fluidity of 30 randomly selected patients using Image J software (version 1.5; National Institutes of Health, Bethesda, MD, USA). To assess intraobserver reproducibility, reader 1 re-read the images and completed the remaining measurements 1 month later.

Lesions were radiologically evaluated for the following features: location (head and neck or body and tail), size (maximum tumor diameters), tumor components, necrosis, intratumoral hemorrhage, apparent diffusion coefficient (ADC) value (b = 800 s/mm² ) and enhancement pattern. The maximum tumor diameter was measured on T2-trigger-turbo-spin echo fat-saturated images [13, 25]. Tumor components were defined as predominantly solid (solid components > 90%), solid-cystic (solid components 10–90%) or predominantly cystic (solid components < 10%) [25]. Tumor necrosis was defined as intratumoral areas showing high-signal intensity on T2WI, and no contrast enhancement on dynamic-enhanced images [26]. The intratumoral hemorrhage signs on the MRI were as follows: (1) high intensity on T1WI (acute-subacute phase) [27], (2) fluid-fluid level in the tumor (chronic phase) [28], or (3) very low intensity of hemosiderin on T2WI (chronic phase) [28]. We evaluated the relative signal intensity of all pancreatic lesions compared with the adjacent normal pancreatic parenchyma and determined the relative enhancement degree of the arterial, portal venous, and delay phases (hypo-, iso-, or hyperintense). The enhancement modes were divided as follows: (1) hyperenhancement in the arterial phase regardless of enhancement degree in the delayed and portal phases [type 1], (2) persistent hypoenhancement in all three phases [type 2], (3) persistent iso-enhancement in all three phases [type 3]. Arterial phase with hypo- or iso-enhancement: (4) portal hyperenhancement with hyperenhancement [type 4a], iso-enhancement [type 4b] or hypoenhancement [type 4c] on delayed-phase images, (5) portal hypoenhancement with gradual delayed hyperenhancement on delayed-phase images [type 5], and (6) portal iso-enhancement with hyperenhancement [type 6a] iso-enhancement [type 6b] or hypoenhancement [type 6c] on delayed-phase images [21]. A clinical worker (reader 3) who was not involved in evaluating the radiological features recorded the clinical-related indicators (e.g., age, sex, symptoms and pathological data).

Statistical analysis

Statistical analysis was performed using SPSS, version 24.0 (SPSS Inc., Chicago, IL, USA), MedCalc software (MedCalc Software, Ltd, version 20.218, Ostend Belgium) and GraphPad Prism (version 5.01, GraphPad Prism for Windows). Continuous variables are expressed as means ± standard deviation or medians (interquartile range) according to the Shapiro-Wilk test and were compared using Student’s t-test or the Kruskal-Wallis test. Categorical data are expressed as frequencies or percentages and analyzed using chi-square or Fisher’s exact tests. Bland-Altman analysis and intraclass correlation coefficients (ICCs) were calculated to assess interobserver agreements in quantitative data, which were graded as follows: 0–0.20 = slight, 0.21–0.40 = fair, 0.41–0.60 = moderate, 0.61–0.80 = substantial, and 0.81–1.00 = near perfect [29]. Qualitative data were calculated using quadratic weighted Cohen’s kappa coefficient (κ): 0–0.40 = slight or fair agreement, 0.41–0.60 = moderate agreement, 0.61–0.80 = substantial agreement, and 0.81–1.00 = near perfect agreement [30]. Receiver operating characteristic (ROC) analysis was conducted, and the optimal threshold of diagnosis was determined using the Youden index. Area under the curve (AUC), sensitivity, and specificity, were calculated. Model performances were compared using Delong’s test. Statistical significance was defined as P < 0.05.

Patient and lesion characteristics

Ninety-two patients were enrolled in the study, 30 with untreated SPNs (9 men, 21 women; mean age, 34.5 ± 12.5 years) and 62 with untreated NF-pNENs (23 men, 39 women; mean age, 48.7 ± 13.9 years). Of the NF-pNENs, 19 were grade 1, 29 were grade 2, and 14 were grade 3 or neuroendocrine carcinomas. Table 2 shows the baseline characteristics of the enrolled patients. SPNs occurred at a younger mean age (34.5 vs 48.7 years, P < 0.001) and clinical symptoms are not as diverse as those of NF-pNENs (P = 0.031). The median c value of SPNs was 1.87 (interquartile range [IQR]:1.77–1.99) m/s, which was lower than that of in NF-pNENs (2.40 m/s, IQR: 2.09–2.86) (P < 0.001). And the median φ value of SPNs was 0.97 (IQR: 0.90–1.04) rad, which was lower than that of in NF-pNENs (1.10 rad, IQR: 1.03–1.19) (P < 0.001) (Table 2 and Fig. 2). The tumor enhancement pattern showed differences between NF-pNENs and SPNs (P = 0.025). Sex, symptoms, tumor location, tumor components, tumor necrosis, intratumoral hemorrhages and ADC values did not significantly differ (P > 0.05). Figure 3 show a clinical cases of NF-pNENs and SPNs.

Table 2

Baseline characteristics of patients with NF-pNENs and SPNs
Variable	NF-pNENs (n = 62)	SPNs (n = 30)	P-value
Age (years)	48.7 ± 13.9	34.5 ± 12.5	< 0.001*
Sex (n, %)			0.662
Male	23 (37)	9 (30)	-
Female	39 (63)	21 (70)	-
Symptoms (n, %)
Asymptomatic	25 (40)	18 (60)	0.121
Abdominal pain or distension	23 (37)	11 (37)	> 0.999
Other symptoms (e.g., vomiting, malaise)	13 (21)	1 (3)	0.031*
Grade of differentiation, n (%)
G1	19 (31)	-	-
G2	29 (47)	-	-
G3 and NEC	14 (22)	-	-
Tumor location			0.961
Head and neck	25 (40)	13 (43)	-
Body and tail	37 (60)	17 (57)	-
Tumor size (mm)	34.5 (24.0, 47.5)	41.5 (32.0, 51.0)	0.141
Tumor components			0.284
Predominant solid	34 (55)	13 (43)	-
Solid-cystic	28 (45)	16 (53)	-
Predominant cystic	0 (0)	1 (3)	-
Necrosis	37 (60)	18 (60)	> 0.999
Hemorrhage	12 (19)	11 (37)	0.123
ADC (×10^− 3 mm²/s)	1015.7 (902.3, 1123.8)	959.7 (852.4, 1114.8)	0.230
Enhancement pattern			0.025*
Type 1	30 (48)	5 (17)	-
Type 2	14 (23)	11 (37)	-
Type 3	4 (6)	1 (3)	-
Type 4 (4a, 4b, 4c)	1 (2)	0 (0)	-
Type 5	8 (13)	7 (23)	-
Type 6 (6a, 6b, 6c)	5 (8)	6 (20)	-
Stiffness (m/s)	2.40 (2.09, 2.86)	1.87 (1.77, 1.99)	< 0.001*
Fluidity (rad)	1.10 (1.03, 1.19)	0.97 (0.90, 1.04)	< 0.001*

NF-pNENs, non-functional pancreatic neuroendocrine neoplasms; SPNs, solid pseudopapillary neoplasms; NEC, neuroendocrine carcinoma; ADC, apparent diffusion coefficient

Continuous variables are presented as means ± standard deviation or median (interquartile range), categorical variables were presented as number (percentage).

^*Statistically significant (P < 0.05).

Interobserver variability of tomoelastography

Bland-Altman analysis showed near-perfect agreements among the measurements between the two readers in stiffness (c) (Fig. 4a, 4b) and fluidity (φ) (Fig. 4c, 4d). The ICCs showed near-perfect agreement in c (0.992, 95% confidence interval [CI]: 0.988–0.995, P < 0.001) and φ (0.908, 95% CI: 0.862–0.940, P < 0.001). The following radiological findings showed substantial or near-perfect interobserver agreement: tumor location (κ = 1.000, P < 0.001), tumor components (κ = 0.834, P < 0.001), necrosis (κ = 0.928, P < 0.001), intratumoral hemorrhage (κ = 0.809, P < 0.001), enhancement pattern (κ = 0.798, P < 0.001), tumor size (ICC = 0.999, 95% CI: 0.999–1.000, P < 0.001) and ADC values (ICC = 0.849, 95% CI: 0.776–0.900, P < 0.001).

Clinical and radiological features for differentiating NF-pNENs and SPNs

Univariate analysis of the clinical and radiological factors revealed that age (P < 0.001; odds ratio [OR], 1.08; 95% CI: 1.04–1.13), enhancement pattern (P = 0.013; OR, 0.74; 95% CI: 0.59–0.94), stiffness (P < 0.001; OR, 1.69; 95% CI: 1.31–2.17) and fluidity (P < 0.001; OR, 1.56; 95% CI: 1.10–2.22) were relevant in differentiating NF-pNENs and SPNs (Table 3). Analyzing only the factors with P < 0.05 in the univariate analysis showed that age (P = 0.002; OR, 1.10;95% CI: 1.03–1.16) and stiffness (P < 0.001; OR, 1.72；95% CI：1.28–2.31) were independent factors in the multivariate regression with forward stepwise analysis (Table 4).

Table 3

Univariate analysis of relevant clinical and radiological factors
Variables	Univariate analysis
Variables	OR	95% CI	P-value
Age (years)	1.08	1.04–1.13	<0.001*
Sex (n, %)	1.38	0.54–3.51	0.456
Symptoms (n, %)
Asymptomatic	0.45	0.19–1.10	0.079
Abdominal pain or distension	1.02	0.41–2.52	0.968
Other symptoms (e.g., vomiting, malaise)	7.69	0.96–61.91	0.055
Tumor location	1.13	0.47–2.74	0.783
Tumor size (mm)	0.99	0.97–1.01	0.432
Tumor components	1.73	0.74–4.01	0.203
Necrosis	0.99	0.41–2.40	0.976
Hemorrhage	0.42	0.16–1.10	0.076
ADC (×10^− 3 mm²/ s)	1.00	1.00–1.00	0.530
Enhancement pattern	0.74	0.59–0.94	0.013*
Stiffness (m/s)	1.69	1.31–2.17	<0.001*
Fluidity (rad)	1.56	1.10–2.22	<0.001*

OR, odds ratio; CI, confidence interval; ADC, apparent diffusion coefficient

^*Statistically significant (P < 0.05).

Table 4

Multivariate analysis of relevant clinical and radiological factors
	Multivariate analysis
	OR	95% CI	P-value
Clinical and MRI parameters
Age	1.08	1.04–1.13	<0.001*
Enhancement pattern	0.76	0.59–0.99	0.041
MRE parameters
Stiffness	1.68	1.29–2.19	<0.001*
Fluidity	1.00	0.72–1.40	0.978
Combined model
Age	1.10	1.03–1.16	0.002*
Enhancement pattern	0.87	0.63–1.20	0.395
Stiffness	1.72	1.28–2.31	<0.001*
Fluidity	0.93	0.70–1.23	0.600

OR, odds ratio; CI，confidence interval

^*Statistically significant (P < 0.05).

Performance of relevant features and models

We separately evaluated the diagnostic performances of four indicators to differentiate NF-pNENs and SPNs. The discrimination performances of stiffness (c) (AUC, 0.876; 95% CI: 0.805–0.947; cutoff, 2.07 m/s) and fluidity (φ) were satisfactory (AUC, 0.788; 95% CI: 0.682–0.894; cutoff, 1.05 rad) and were both higher than that of age (AUC, 0.780; 95% CI: 0.680–0.879; cutoff, 47.5 years) and enhancement pattern (AUC, 0.675; 95% CI: 0.564–0.787). We combined the two independent factors in the multivariate analysis into a combined model (c + age). And combined the age and enhancement pattern into a conventional model (age + enhancement pattern). Performance of the combined model (c + age) was excellent (AUC, 0.921; 95% CI: 0.865–0.977), and the difference was significant compared with conventional model (AUC, 0.813;95% CI: 0.717–0.909, P=0.009), and higher than age (P < 0.001), enhancement pattern (P < 0.001) and fluidity (φ) (P = 0.011). The combined model (c + age) performed slightly better than using stiffness (c) (AUC = 0.921 vs 0.876) alone, but not significantly (P = 0.115; Table 5 and Fig. 5).

Table 5

Performances of the models for differentiating NF-pNENs and SPNs
	Delong’s Test	AUC (95% CI)	Ac	Se	Sp	PPV	NPV	Cutoff value
Age	<0.001*	0.780 (0.680–0.879)	0.674	0.565	0.900	0.921	0.500	47.5
Enhancement pattern	<0.001*	0.675 (0.564–0.787)	0.598	0.484	0.833	0.857	0.439	/
c	0.115	0.876 (0.805–0.947)	0.826	0.774	0.933	0.960	0.667	2.07
φ	0.011*	0.788 (0.682–0.894)	0.739	0.710	0.800	0.880	0.571	1.05
Age + enhancement pattern	0.009*	0.813 (0.717–0.909)	0.804	0.871	0.667	0.844	0.714	/
c + Age	-	0.921 (0.865–0.977)	0.848	0.790	0.967	0.980	0.690	/

AUC, area under the curve; CI, confidence interval; Ac, accuracy; Se, sensitivity; Sp, specificity; PPV, positive predictive value; NPV, negative predictive value

^*Statistically significant (P < 0.05).

NF-pNENs are clinically similar to SPNs and overlap in imaging features; however, their prognoses and treatment strategies differ [14, 25]. In clinical precision medicine, a noninvasive modality that can differentiate NF-pNENs and SPNs before surgery is urgently needed. We found that stiffness (c), fluidity (φ), age and enhancement pattern were relevant factors in distinguishing these tumors. But only stiffness (c) and age were independent factors. The performances of combined model (c + age) was satisfactory and better than that of conventional model (age + enhancement) in differentiating NF-pNENs and SPNs, but comparable to use stiffness (c) alone. To our knowledge, this is the first report on multifrequency MRE using stiffness and fluidity to differentiate NF-pNENs and SPNs.

Accumulating evidence suggests that the mechanical properties of tumors provide valuable information for diagnosis and differentiation [31].Tomoelastography, a novel and promising technology, provides high-resolution parameters such as stiffness and fluidity, which enable quantitatively evaluating the mechanical characteristics of pancreatic tumors. Compared with traditional elastography, tomoelastography is better for studying small abdominal and retroperitoneal organs [18]. We found that higher stiffness and fluidity helped in differentially diagnosing NF-pNENs and SPNs. This is similar to the findings of Gültekin et al. who characterized pNENs aggressiveness based on stiffness [32]. Stiffness is related to the degree of fibrosis [33]. Previous studies demonstrated that NF-pNENs are promoted by many growth factors and exhibit varying levels of fibrosis [34]. Funa et al. found increased expression of the PDGF-b receptor in stromal cells adjacent to tumor cells primarily in midgut and pancreatic neuroendocrine tumors, which may contribute to significant fibrosis around tumor tissue [35]. Conversely, one characteristic of SPNs is their widely cystic regions, and reduction of solid structural components can lead to decreased stiffness.

Fluidity is a relatively less researched parameter. Streitberger et al. reported that fluidity reflects only the internal mechanical friction of biological tissue rather than the total water content [20]. Previous studies have found that fluidity is correlated with the invasiveness of malignant tumors because it reflects cellular adhesion in cell-rich tissues [22, 23, 36]. More invasive tumors exhibit higher fluidity values. NF-pNENs have more malignant tumor characteristics than do SPNs, enabling NF-pNENs to accumulate more hydrophobic and disorganized proteins, leading to increased stiffness and fluidity by increasing tissue stiffness and mechanical friction [22]. Furthermore, NF-pNENs exhibit hypervascular characteristics [37]. Gültekin et al. believed this feature helps increase tumor stiffness [32], whereas Shahryari et al. reported that it may lead to relatively high fluidity in malignant tumors [22]. These mechanical changes may enable distinguishing NF-pNENs from SPNs.

We found that patient age and tumor size or components differed between NF-pNENs and SPNs, but only age was an independent factor in the diagnosis. Previous studies showed that SPNs occurred mostly in patients aged 33–34 years[25, 38], where as NF-pNENs were more common in patients aged approximately 50 years [39]. In our study, the mean patient ages were 34.5 years for SPNs and 48.7 years for NF-pNENs, which is highly consistent with previous studies. Consistent with our findings, many studies have reported that SPNs are larger than NF-pNENs [4, 14, 40]. In contrast to previous results [40], we found no differences between sexes in our study. Shao et al. and Dogan et al. found that NF-pNENs were more common in women, similar to the SPNs [41]. Thus, we believe that age may be more advantageous than sex for distinguishing SPNs from NF-pNENs, as both have certain female preferences.

Previous studies have reported that SPNs exhibit slight contrast enhancement in the arterial phase and progressive enhancement in other phases, whereas the enhancement pattern in NF-pNENs is correlated with tumor grade and differentiation. Well-differentiated NF-pNENs exhibit hyperenhancement patterns, whereas poorly differentiated NF-pNENs, such as grades 3 neuroendocrine tumors and NEC, are correlated with hypoenhancement patterns [42, 43]. In our study, there were differences in enhancement patterns between NF-pNENs and SPNs, which is similar to previous studies[25].

This study had several limitations. First, it was a single-institutional study conducted in a large tertiary hospital. Second, the sample sizes were relatively small because many patients had already undergone treatment and were thus excluded. Therefore, a much larger database with multicenter validation and a larger sample is needed to confirm our results. Finally, our data were acquired from a single manufacturer of 3T MRI. Whether different MRI field strengths and vendors would affect the results requires further study.

Tomoelastography is a noninvasive and novel MR technology that is beneficial for quantitatively analyzing pancreatic diseases. Stiffness combined with clinical age metrics were effective in identifying NF-pNENs and SPNs. The differentiation performance was improved compared to the conventional model (age + enhancement pattern), establishing the value of tomoelastography in the non-invasive preoperative quantitative identification of pancreas-associated neoplasms.

pNENs: pancreatic neuroendocrine neoplasms

NF-pNENs: non-functional pancreatic neuroendocrine neoplasms

SPNs: solid pseudopapillary neoplasms

MRE: magnetic resonance elastography

ROI: region of interest

AUC: area under the curve

ADC: apparent diffusion coefficient

DWI: diffusion-weighted imaging

CI: confidence interval

Ethics approval and consent to participate: The ethics committee of the first affiliated hospital of Sun Yat-sen University institution approved this prospective single-center study and written informed consent was obtained from all the patients.(approval number: 2021 [721]).

Consent for publication: All presentations of case reports have been consent for publication.

Availability of data and materials: The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests: The authors of this manuscript declare no relationships with any companies, whose products or services may be related to the subject matter of the article.

Funding: This study has received funding from the National Natural Science Foundation of China (81801761 to Yanji Luo) and Natural Science Foundation of Guangdong Province (2024A1515011968 to Yanji Luo).

Authors’ contributions: All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Jiaxin Yuan, Jinhui Yu, Siya Shi, Liqin Wang and Tingting Wen. The first draft of the manuscript was written by Jiaxin Yuan and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Acknowledgments: The authors sincerely acknowledge Ms. Jing Guo from the Department of Radiology of Charité-Universitätsmedizin Berlin, Germany, for the MR technical support.

Cives M, Strosberg J R. Gastroenteropancreatic Neuroendocrine Tumors. CA Cancer J Clin 2018; 68(6): 471-487.
Liu Y, Shi S, Hua J, et al. Differentiation of solid-pseudopapillary tumors of the pancreas from pancreatic neuroendocrine tumors by using endoscopic ultrasound. Clin Res Hepatol Gastroenterol 2020; 44(6): 947-953.
Abdelkader A, Hunt B, Hartley C P, Panarelli N C, Giorgadze T. Cystic Lesions of the Pancreas: Differential Diagnosis and Cytologic-Histologic Correlation. Arch Pathol Lab Med 2020; 144(1): 47-61.
Kataoka K, Ishikawa T, Ohno E, et al. Differentiation Between Solid Pseudopapillary Neoplasm of the Pancreas and Nonfunctional Pancreatic Neuroendocrine Neoplasm Using Endoscopic Ultrasound. Pancreas 2022; 51(1): 106-111.
Pavel M, Oberg K, Falconi M, et al. Gastroenteropancreatic neuroendocrine neoplasms: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol 2020; 31(7): 844-860.
Partelli S, Bartsch D K, Capdevila J, et al. ENETS Consensus Guidelines for the Standards of Care in Neuroendocrine Tumours: Surgery for Small Intestinal and Pancreatic Neuroendocrine Tumours. Neuroendocrinology 2017; 105(3): 255-265.
Cherif R, Gaujoux S, Couvelard A, et al. Parenchyma-Sparing Resections for Pancreatic Neuroendocrine Tumors. Journal of Gastrointestinal Surgery 2012; 16(11): 2045-2055.
Yamao K. Complications of endoscopic ultrasound-guided fine-needle aspiration biopsy (EUS-FNAB) for pancreatic lesions. Journal of Gastroenterology 2005; 40(9): 921-923.
Lindholm D P, Oberg K. Biomarkers and molecular imaging in gastroenteropancreatic neuroendocrine tumors. Horm Metab Res 2011; 43(12): 832-837.
Marotta V, Zatelli M C, Sciammarella C, et al. Chromogranin A as circulating marker for diagnosis and management of neuroendocrine neoplasms: more flaws than fame. Endocr Relat Cancer 2018; 25(1): R11-R29.
Baudin E, Gigliotti A, Ducreux M, et al. Neuron-specific enolase and chromogranin A as markers of neuroendocrine tumours. Br J Cancer 1998; 78(8): 1102-1107.
Nobels F R, Kwekkeboom D J, Coopmans W, et al. Chromogranin A as serum marker for neuroendocrine neoplasia: comparison with neuron-specific enolase and the alpha-subunit of glycoprotein hormones. J Clin Endocrinol Metab 1997; 82(8): 2622-2628.
Shi Y J, Li X T, Zhang X Y, et al. Non-gaussian models of 3-Tesla diffusion-weighted MRI for the differentiation of pancreatic ductal adenocarcinomas from neuroendocrine tumors and solid pseudopapillary neoplasms. Magn Reson Imaging 2021; 83: 68-76.
Liu B A, Li Z M, Su Z S, She X L. Pathological differential diagnosis of solid-pseudopapillary neoplasm and endocrine tumors of the pancreas. World J Gastroenterol 2010; 16(8): 1025-1030.
Dinarvand P, Lai J. Solid Pseudopapillary Neoplasm of the Pancreas: A Rare Entity With Unique Features. Arch Pathol Lab Med 2017; 141(7): 990-995.
McCall C M, Shi C, Klein A P, et al. Serotonin expression in pancreatic neuroendocrine tumors correlates with a trabecular histologic pattern and large duct involvement. Hum Pathol 2012; 43(8): 1169-1176.
Johnson A, Wright J P, Zhao Z, et al. Cadherin 17 is frequently expressed by 'sclerosing variant' pancreatic neuroendocrine tumour. Histopathology 2015; 66(2): 225-233.
Tzschätzsch H, Guo J, Dittmann F, et al. Tomoelastography by multifrequency wave number recovery from time-harmonic propagating shear waves. Medical Image Analysis 2016; 30: 1-10.
Venkatesh S K, Wells M L, Miller F H, et al. Magnetic resonance elastography: beyond liver fibrosis-a case-based pictorial review. Abdom Radiol (NY) 2018; 43(7): 1590-1611.
Streitberger K J, Lilaj L, Schrank F, et al. How tissue fluidity influences brain tumor progression. Proc Natl Acad Sci U S A 2020; 117(1): 128-134.
Jeon S K, Lee J M, Joo I, et al. Nonhypervascular Pancreatic Neuroendocrine Tumors: Differential Diagnosis from Pancreatic Ductal Adenocarcinomas at MR Imaging—Retrospective Cross-sectional Study. Radiology 2017; 284(1): 77-87.
Shahryari M, Tzschatzsch H, Guo J, et al. Tomoelastography Distinguishes Noninvasively between Benign and Malignant Liver Lesions. Cancer Res 2019; 79(22): 5704-5710.
Marticorena Garcia S R, Zhu L, Gultekin E, et al. Tomoelastography for Measurement of Tumor Volume Related to Tissue Stiffness in Pancreatic Ductal Adenocarcinomas. Invest Radiol 2020; 55(12): 769-774.
Shi S Y, Wang L, Peng Z, et al. Multi-frequency magnetic resonance elastography of the pancreas: measurement reproducibility and variance among healthy volunteers. Gastroenterol Rep (Oxf) 2022; 10: goac033.
Wang C, Cui W, Wang J, Chen X, Tong H, Wang Z. Differentiation between solid pseudopapillary neoplasm of the pancreas and hypovascular pancreatic neuroendocrine tumors by using computed tomography. Acta Radiol 2019; 60(10): 1216-1223.
Kim H, Kim D H, Song I H, et al. Survival Prediction after Curative Resection of Pancreatic Ductal Adenocarcinoma by Imaging-Based Intratumoral Necrosis. Cancers (Basel) 2022; 14(22).
Bonneville F, Cattin F, Marsot-Dupuch K, Dormont D, Bonneville J F, Chiras J. T1 signal hyperintensity in the sellar region: spectrum of findings. Radiographics 2006; 26(1): 93-113.
Kurihara N, Takahashi S, Higano S, et al. Hemorrhage in pituitary adenoma: correlation of MR imaging with operative findings. Eur Radiol 1998; 8(6): 971-976.
Min J H, Lee M W, Park H S, et al. Interobserver Variability and Diagnostic Performance of Gadoxetic Acid-enhanced MRI for Predicting Microvascular Invasion in Hepatocellular Carcinoma. Radiology 2020; 297(3): 573-581.
Kim J H, Eun H W, Kim Y J, Lee J M, Han J K, Choi B I. Pancreatic neuroendocrine tumour (PNET): Staging accuracy of MDCT and its diagnostic performance for the differentiation of PNET with uncommon CT findings from pancreatic adenocarcinoma. Eur Radiol 2016; 26(5): 1338-1347.
Paszek M J, Zahir N, Johnson K R, et al. Tensional homeostasis and the malignant phenotype. Cancer Cell 2005; 8(3): 241-254.
Gultekin E, Wetz C, Braun J, et al. Added Value of Tomoelastography for Characterization of Pancreatic Neuroendocrine Tumor Aggressiveness Based on Stiffness. Cancers (Basel) 2021; 13(20).
Reiter R, Tzschatzsch H, Schwahofer F, et al. Diagnostic performance of tomoelastography of the liver and spleen for staging hepatic fibrosis. Eur Radiol 2020; 30(3): 1719-1729.
Laskaratos F M, Rombouts K, Caplin M, Toumpanakis C, Thirlwell C, Mandair D. Neuroendocrine tumors and fibrosis: An unsolved mystery? Cancer 2017; 123(24): 4770-4790.
Funa K, Papanicolaou V, Juhlin C, et al. Expression of platelet-derived growth factor beta-receptors on stromal tissue cells in human carcinoid tumors. Cancer Res 1990; 50(3): 748-753.
Oswald L, Grosser S, Smith D M, Kas J A. Jamming transitions in cancer. J Phys D Appl Phys 2017; 50(48): 483001.
van Essen M, Sundin A, Krenning E P, Kwekkeboom D J. Neuroendocrine tumours: the role of imaging for diagnosis and therapy. Nat Rev Endocrinol 2014; 10(2): 102-114.
Xu Y, Zhao G, Pu N, et al. One Hundred Twenty-One Resected Solid Pseudopapillary Tumors of the Pancreas: An 8-Year Single-Institution Experience at Zhongshan Hospital, Shanghai, China. Pancreas 2017; 46(8): 1023-1028.
Low G, Panu A, Millo N, Leen E. Multimodality imaging of neoplastic and nonneoplastic solid lesions of the pancreas. Radiographics 2011; 31(4): 993-1015.
Song T, Zhang Q W, Duan S F, et al. MRI-based radiomics approach for differentiation of hypovascular non-functional pancreatic neuroendocrine tumors and solid pseudopapillary neoplasms of the pancreas. BMC Med Imaging 2021; 21(1): 36.
Shao S, Zeng Z, Hu S. An observational analysis of insulinoma from a single institution. QJM 2018; 111(4): 237-241.
Li X, Zhu H, Qian X, Chen N, Lin X. MRI Texture Analysis for Differentiating Nonfunctional Pancreatic Neuroendocrine Neoplasms From Solid Pseudopapillary Neoplasms of the Pancreas. Academic Radiology 2020; 27(6): 815-823.
Kim D W, Kim H J, Kim K W, et al. Neuroendocrine neoplasms of the pancreas at dynamic enhanced CT: comparison between grade 3 neuroendocrine carcinoma and grade 1/2 neuroendocrine tumour. European Radiology 2014; 25(5): 1375-1383.

No competing interests reported.

Download PDF

Editor invited by journal
23 Aug, 2024
Reviewers invited by journal
16 Apr, 2024
Editor assigned by journal
16 Apr, 2024
Submission checks completed at journal
15 Apr, 2024
First submitted to journal
14 Apr, 2024

You are reading this latest preprint version

Role of tomoelastography in differentiating between non-functional pancreatic neuroendocrine neoplasms and solid pseudopapillary neoplasms

Status:

Version 1

Abstract

Background

Methods

Results

Conclusions

Figures

Background

Methods

Study population

MRI and multifrequency MRE

Tomoelastography postprocessing

Image analysis

Statistical analysis

Results

Patient and lesion characteristics

Interobserver variability of tomoelastography

Clinical and radiological features for differentiating NF-pNENs and SPNs

Performance of relevant features and models

Discussion

Conclusions

Abbreviations

Declarations

References

Additional Declarations

Status:

Version 1