Key words
insulinoma - insulin - endoscopic ultrasound - imaging - 72-hour fast - hypoglycemia - tumor size
Introduction
Insulinoma is a rare tumor of the endocrine pancreas with an incidence of
1–4/100.000 per year [1].
It can cause spontaneous hypoglycemia due to excessive insulin production [2]. Women are more likely to get an
insulinoma [3], and there is a peak of
manifestation in the 5th decade of life [4]. Mostly, insulinomas are small solitary
lesions, that are benign [5]. The 72-hour
fast is the gold standard diagnostic test for recreating the conditions that lead to
hypoglycemia [6]. Surgical resection is
the preferred therapy because of its curative intention [7]. Preoperative precise localization is
mandatory but often difficult [8]. The
most commonly used techniques for localization include transabdominal ultrasound
(US), computer tomography (CT), and magnetic resonance imaging
(MRI). All listed modalities are able to detect insulinomas with a
sensitivity up to 80% [9]
.
Endoscopic ultrasound (EUS) is another established method to find the
suspicious lesion. Although invasive, it has little to no harm to the patient and is
the most sensitive method to localize this pancreatic tumor [10]. Although established diagnostic
routes and criteria lead to correct results, there is often a delay in making an
accurate diagnosis. Reasons for this are unspecific symptoms, which range from
neuroglycopenic features to autonomic errors [11]. The reason that neuroendocrine tumors are slow-growing neoplasms and
liberate insulin and proinsulin in secretory bursts [12] aggravates these pitfalls. Only a few
publications have investigated the correlation between the size of the tumor, their
characteristics, and the duration of the 72-hour fast [13]
[14]. This study aimed to survey whether the measured size by EUS, the
echogenic texture, and secretory characteristics of insulinoma correlate with the
duration of the fasting test and hence the severity of the disease.
Methods
This retrospective study included 45 patients diagnosed with insulinoma. Each tumor
was examined endosonographically by one single investigator (phk) in the
university hospitals in Mainz and Marburg, Germany. Endosonographic ultrasound was
performed with Pentax FG32UA and FG36UX endoscopes (Pentax cooperation, Tokyo,
Japan) with a longitudinal 7.5 MHz sector array combined with a Hitachi EUB
420 or Hitachi EUB 525 ultrasound computer (Hitachi medical cooperation, Tokyo,
Japan).
The inclusion criteria were:
-
Positive 72-hour fast due to biochemical and clinical values [11]
-
Cyto-/histopathological confirmed insulinoma
-
Available sizes of insulinoma, measured by endoscopic ultrasound and
description of morphology by EUS
-
Available and complete data on insulin, C-peptide, proinsulin, and duration
of 72-hour fasting
-
No relapse of an insulinoma
With the help of two diameters (longest and the diameter at 90°) from the
EUS and the mean calculated value of these, a spherical volume was
defined using the following formula:
V=1/6*π*d3.
Insulin and C-peptide were tested by CLIA-Test (chemiluminescence immunoassay,
Elecsys Insulin, and Elecsys C-Peptide, Roche Diagnostics GmbH, Mannheim, Germany)
and proinsulin by an enzyme-linked immunosorbent assay (ELISA)-test kit
(Immundiagnostik AG, Bensheim, Germany).
For the 72-hour fast, we used the absolute time from the beginning until diagnostic
criteria for discontinuing were met ([Table
1]), and hormone levels (insulin, C-peptide, and proinsulin) were
documented at this endpoint.
Table 1 The diagnostic criteria for the 72-hour fast for the
detection of endogenous hyperinsulinism (adapted from [6]
[40]).
|
Plasma glucose
|
≤45 mg/dL (2.5 mmol/L)
|
|
Insulin
|
≥6 μU/mL (≥36 pmol/L)
|
|
C-peptide
|
≥0.6 μg/L (≥2 nmol/L)
|
|
Proinsulin
|
≥5 pmol/L
|
|
ß-hydroxybutyrate
|
≤2.7 mmol/L
|
|
No evidence for exogeneous insulin or insulin secretagogues
|
|
We calculated the area under the curve of insulin (AUCinsulin) by applying the
trapezoid rule from a 5-hour oral glucose tolerance test (OGTT) with 75 g of
glucose. This data was collected from 16 patients. Furthermore, we plotted and
visualized the mean values for insulin and glucose over the time measured ([Fig. 1]).
Fig. 1 The mean curves of insulin and glucose after glucose exposure
with 75 g oral glucose with standard deviation.; The mean value
curves for insulin and glucose showed a parallel course. However, the
insulin values indicate a high standard deviation, assuming a strong
heterogeneity of the secretion behavior. Furthermore, 50% of the
patients had hypoglycemia after glucose exposure.
Statistical analysis was performed with jamovi software (version 1.6.18) and
Microsoft Excel for Mac (version 16.29).
We used univariant linear regression to determine the association between the
duration of the 72-hour fast, insulin, C-peptide, proinsulin levels, and the
AUCinsulin (subgroup analysis of 16 patients) and the volume measured by EUS. Due to
the non-existent normal distribution of the residuals, we used the bootstrapping
method [15]. This analysis was made with
the help of SPSS (version 2020 for Mac). We used 2000 samples for
bootstrapping with a 95% bias-corrected and accelerated
confidence interval.
In addition, we recorded specific properties of echotexture by means of EUS, based on
the experience and assessment of the examiner ([Fig. 2 a,b]).
Fig. 2
a EUS image of a homogenous insulinoma (arrow) located in the
pancreatic head; ventral to the superior mesenteric-portal venous (SMPV)
confluence (*). b EUS image of a non-homogenous insulinoma
(arrow) located in the pancreatic head; the tumor has a hyperechoic margin
in the form of a fat capsule; ventral to the superior mesenteric-portal
venous (SMPV) confluence (*).; volume
(cm
3
); Mean 2.3; Median 1.2; Standard
deviation 2.9; Minimum 0.1; Maximum 14.2.
We divided the study population into two groups: homogenous and non-homogenous
echogenicity [16].
The differences between homogenous tumors and non-homogenic ones were assessed in
terms of their size, duration of the 72-hour fast, Ki-67 index, and hormone
levels.
For comparisons between these two, the Mann-Whitney U test was used.
Statistical significance was assumed at a P-value of<0.05.
Results
We examined 28 females (62.2%) and 17 males (37.8%), with a mean age
at the time of diagnosis of 47 ± 17 years. The insulinomas were distributed
in the pancreatic caput (57.8%), in the corpus (17.8%), and in the
cauda (24.4%). All insulinomas were confirmed by
cyto-/histopathological analysis. Just one insulinoma was classified as
malignant because of lymphogenic metastasis, and five patients were suffering from
multiple endocrine neoplasia 1 (MEN 1).
Most insulinomas were small, with a median of 1.15 cm3 ([Fig. 3]). The median duration of the
72-hour fast was 10 hours, with one test that lasted over the entire
duration of 72 hours. At the end of the 72-hour fast, the hormone levels
were widely distributed ([Table 2]).
Fig. 3 The distribution of the volumes measured.
Table 2 The hormone-levels measured at the end of the
fast.
|
Insulin mULl
|
C-peptide μg/L
|
Proinsulin pmol/L
|
|
Mean
|
23.6
|
3.2
|
90.4
|
|
Median
|
14.7
|
2.6
|
40.3
|
|
Standard deviation
|
25.3
|
2.1
|
198
|
|
Minimum
|
2.2
|
0.3
|
5.70
|
|
Maximum
|
124
|
9.7
|
1059
|
No significant relationship was observed between the size of insulinoma measured
endosonographically, the duration of a 72-hour fast, and the measured hormone
levels. The AUCinsulin of the 5-hour OGTT also showed no correlation with the size
of the insulinomas. The curves of the mean values of insulin and glucose showed an
almost parallel progression, although the standard deviations were very large ([Fig. 1]).
The endosonographic assessment of the tumor texture revealed that 31 (69%)
had a homogeneous echotexture, whereas the remaining 14 were non-homogeneous.
Non-homogeneous insulinoma were significantly larger and had a higher Ki-67 index,
ranging from G1 (≤2%) to G2 (3–20%). There was no
statistically substantial difference in the other values considered, not even in the
subgroup analysis related to the AUCinsulin. However, each median of the hormone
levels was higher in the homogeneous group, although not statistically significant.
In the case of homogenous tumors, there was a tendency toward a shorter duration of
the 72-hour fast ([Table 3]).
Table 3 Comparison between homogenous and non-homogenous
tumors with regard to the duration of the fast, hormone levels, and
Ki-67 index (medians).
|
Homogenous
|
Non-homogenous
|
P-value*
|
|
n
|
31
|
14
|
|
|
Volume (cm3)
|
1.02
|
2.69
|
0.007
|
|
Duration (h)
|
10
|
12.5
|
0.197
|
|
Insulin (mU/l)
|
15.9
|
13.3
|
0.508
|
|
C-peptide (μg/L)
|
2.7
|
2.55
|
0.975
|
|
Proinsulin (pmol/l)
|
44.2
|
36.3
|
1
|
|
Ki-67 (%)
|
1
|
2
|
0.01
|
|
subgroup analysis AUCinsulin
|
|
|
|
N
|
7
|
9
|
|
|
AUCinsulin (mU/L* min)
|
13729
|
15267
|
0.681
|
Discussion
Neuroendocrine tumors cause symptoms due to the autonomic secretion of hormones that
lead to specific symptoms such as spontaneous hypoglycemia.
Although studies of the sizes and biochemical behavior of these lesions are common
[17]
[18], only a few investigations have
correlated symptoms and size. This study has shown that there is no association
between tumor size and severity of the disease by comparing parameters using
endosonographic insulinoma measurements and description of echotexture and the
characteristics of the 72-hour fast as the gold standard of diagnosis. The absolute
number of hours this test lasted is probably the best marker for the patient
discomfort and the severity of the disease.
The distribution of age, sex, and the fact that most insulinomas were small, show
that the examined cohort in this study is representative [19]. Even though this cohort had some
large tumors, just one insulinoma was classified as malignant due to metastases.
This somehow contradicts the description in current guidelines [7]. However, some case reports have also
been published about giant insulinoma that demonstrate the same benign behavior
[20]
[21].
EUS is often found to be the most sensitive localization technique for the detection
of neuroendocrine tumors of the pancreas [22]. Due to its dynamic mode, the investigator can examine the tumor from
different angles and positions so that the size, in particular of small tumors
(<20 mm) can be precisely determined. In addition, EUS shows an
“in vivo” picture of the lesion, since the blood supply to the tumor
is interrupted after the surgical resection, with the possibility that the
pathologist may have a smaller diameter as a result [23]. The Center for Endocrinology,
Diabetology & Osteology of Philipps-University Marburg has experience with
the application of this method, which has been performed by one physician (phk) for
decades. Self-reported data has shown that the detection rate is up to 88%
[24].
Attempts in the past to show a correlation between tumor size and the outcome of a
72-hour fasting gave mixed results. Wolf et al. 2015 showed a positive relationship
between Ki-67 index and hormone levels, a shorter fasting test, and the size of
insulinoma. However, they used modified cut-off values for the Ki-67 index [14]. Two years later, Donegan et al.
published a study that found a relationship between tumor size and the duration of a
72-hour fast [13]. They used a maximum
diameter and no further description of, the texture of the tumor, for example.
Our data suggest that there is no relationship between the volume and the clinical
and biochemical behavior of the tumor. This corresponds with studies on other
neuroendocrine tumors, for example, ACTH-secreting pituitary adenomas [25]. We found out that homogenous
insulinomas are usually small and there is an indication for a shorter fasting
period; on the other hand, bigger tumors have a non-homogenous echotexture and are
not so hyperfunctioning, independent of the KI-67 index. Similar results were shown
by Buetow et al., who described that large tumors are mostly cystic and necrotic and
not hyperfunctioning [26]. Insulinomas
represent a heterogenous group of tumors that show a different secretory behavior,
for example, due to changes in the growth behavior of beta cells, abnormalities in
the somatostatin receptor expression, or the exocytosis of the insulin granules
[27]. In addition, the presence of
hexokinase I enables the secretion of insulin at low blood sugar levels in some
insulinomas [28]. Boden et al. put
forward the theory that insulinomas mainly show glucose transporter type 1 (GLUT1)
expression and healthy beta cells, mainly GLUT2 [29]. By doing this, they explained that
insulinomas can secrete insulin even when glucose levels are low. However, due to
more recent findings, this must be viewed critically, as GLUT2, in contrast to GLUT1
and 3, play a rather subordinate role in humans [30]. The evaluation of the OGTT over
5 hours also showed that 50% of the patients developed hypoglycemia
after the glucose exposure. This also confirms the different secretion profiles of
the insulinomas examined. This is in line with several past observations [31]
[32]. Wiesli et al. 2004 described two patients with insulinoma who had a
negative 72-hour fast but exhibited hypoglycemia after glucose exposure. They
concluded that, in these cases, it was not an incorrect glucose sensing that was
responsible for hypoglycemia but rather an excessive insulin response to glucose
exposure [32].
Due to this heterogenicity, the correlation of the hormone values can only be used to
a limited extent because not only the amount of insulin released is responsible for
the hypoglycemia, but also insulin is released at all when glucose levels are low
[33]. Since insulin is known to
inhibit counter-regulatory hormones, especially glucagon, this leads primarily to
hypoglycemia [34]
[35]
[36]. This is also reflected by the wide range of insulin, C-peptide, and
proinsulin measured that all led to hypoglycemia, as well as the different secretion
patterns and amounts of insulin that followed a glucose stimulus.
Based on these findings, it seems difficult to establish a connection between the
size of the insulinomas and their secretory behavior because this depends on the
type of cells and their potential to secrete insulin and not primarily on their
number. Furthermore, we are aware that the duration of the 72-fast test depends not
only on the insulin secretion but also largely on the interaction of the
contra-insulin hormones such as glucagon, cortisol, catecholamines, and growth
hormone [37]. This was not taken into
account here. Furthermore, we have to note that, to the best of our knowledge, there
are no comparable data on the AUCinsulin of the OGTT over 5 hours. These
measurements are used in diabetes research [38]. Here, the OGTT insulin values are usually used over 2 hours
so that the values cannot be compared with each other so that only an
inter-individual comparison could be made in our study.
Another limitation of this study is that while the EUS is the most sensitive
diagnostic method for neuroendocrine pancreatic tumors, it is also the most
subjective. The assessment of the homogeneity was determined at the discretion of
the investigator. As far as we know, there is no scaling for this, but the
evaluation of whether a tumor is homogeneous or not has already been published by
phk in scientific publications [39].
Since this is a retrospective work and no causality can be established on the basis
of observation alone, this is a purely descriptive work.
Nevertheless, we are convinced that this investigation supports the clinical work, as
we have often observed in our clinical practice that endogenous hyperinsulinism has
been proven in laboratory tests, but non-invasive methods could not be used to
detect insulinoma. This data supports the importance of further, perhaps also
invasive, diagnostics (e. g., EUS) since small insulinomas, in particular,
are difficult to identify.
In conclusion, this study contributes to answering the following question:
“Does the size of the insulinoma matter?”. According to our results,
we can postulate that there is no connection between the measured volume and the
characteristics of the 72-hour fast. Echogenicity seems to be more important than
size for the expression of symptoms. These tumors are possibly well-differentiated
without necrotic areas that lead to a higher and continuous insulin secretion.
Data Availability Statement
Data Availability Statement
The data that supports the results (findings) of this study are available from the
corresponding author upon reasonable request.
