- Top of page
- Conflict of interest
Although the biochemical diagnosis of insulinomas is straightforward, the clinical challenge lies in localization of the tumour within the pancreas. Preoperative localization is essential to increase the chances of surgical success, minimize operating room time, reduce the rate of recurrence and avoid unnecessary resection of the pancreas.[1, 2] A number of noninvasive localization techniques are available including ultrasonography (US), computed tomography (CT), magnetic resonance imaging (MRI) and photon emission tomography (PET), as well as older invasive procedures such as pancreatic angiography and transhepatic portal venous sampling. In general, the sensitivity of these imaging techniques has been reported to be low. Furthermore, despite advances in the technology of scanners, a recent study by the National Institute of Health found that there has been no improvement in the sensitivity of present-day noninvasive imaging tests for localization of insulinomas compared with those available two decades ago. More recent invasive localization modalities include endoscopic US (EUS), intraoperative US and selective intra-arterial calcium stimulation of insulin release by injection into the major pancreatic arteries with hepatic venous sampling for serum insulin (SACST). Although these have all had generally superior detection rates compared with non-invasive scans, SACST has been particularly successful, with reported successful localization rates of 67–100%.[4-8]
Doppman and co-workers originally described the SACST technique in 1991 using calcium doses of 0·01–0·025 mEq/kg. This has been successfully repeated by numerous other studies using the 0·025 mEq/kg dosage.[3-8] In an effort to lower the dose of calcium required and reduce the risk of hypoglycaemia, O'Shea et al. titrated the dose of calcium gluconate to 0·00625 mEq/kg, which they found to be equally as effective as the 0·025 mEq/kg dose used by Doppman et al. In the present study, the sensitivity of the SACST technique has been studied using two low doses of calcium gluconate, 0·00625 mEq/kg and 0·0025 mEq/kg.
- Top of page
- Conflict of interest
The procedure was well tolerated with no significant complications. Most individuals experienced minor flushing with each calcium infusion, but this was not troublesome. The intensity of flushing was not graded in a systematic fashion, but the nurses involved in the procedure noted that flushing and tingling was milder in the very low-dose calcium group (protocol A) than in protocol B. No patient experienced hypoglycaemia.
The definition of a positive response to calcium infusion in a given artery has not been rigorously defined in the literature on the SACST procedure. Traditionally, a twofold rise in serum insulin concentration has been used to define a sufficient increase to indicate localization of an insulinoma perfused by the injected artery. In order to estimate the increment in insulin production that would indicate a positive response using the current calcium dosages, the individual SACST test results were reviewed to compare the fold step-up in hepatic vein insulin concentration following injection into arteries at the opposite end of the pancreas from where the tumour was detected surgically in those with histologically confirmed tumours (e.g. the splenic artery for patients who had insulinomas in the head of the pancreas, or the gastroduodenal artery or hepatic artery for those with tumours in the pancreatic tail). The insulin increments following calcium injection into arteries in these regions are shown in Table 2 and were used to provide control data for noninsulinoma islet tissue. For Protocol A, the mean insulin increment compared with baseline in noninvolved regions was 1·0 (range 0·7–1·2) for patients with proven insulinomas (patients 1–10), and for Protocol B was 1·1 (range 0·5–1·5) for patients 12–17 with proven insulinomas. A positive response to calcium infusion was thus defined as a step-up in insulin concentration of >1·3 for Protocol A patients and >1·5 for the Protocol B group.
Table 2. Results of localization tests and surgery
|Patient no||Pre-SACST scans and localizationa||Artery with maximum insulin increaseb||Artery with minimal insulin increasec||SACST localization||Site of tumour at surgery||Histology|
|Group A: Calcium infusion 0·0025 mEq/kg|
| 1||C = neg||Hepatic (6·2)||Distal splenic (1)||Head||Head||Insulinoma|
| 2||C = neg. U = neck||SMA (2·2)||Distal splenic (0·9)||Inferior head||Inferior head||Insulinoma|
| 3||Nil||Gastroduodenal (2·5)||Distal splenic (1)||Head||Anterior head||Insulinoma|
| 4||Nil||Distal splenic (8·4)||Hepatic (0·8)||Body/tail||Tail||Insulinoma|
| 5||C = body, U = body||Coeliac (5·5)||Distal splenic (0·9)||Head||Superior neck||Insulinoma|
| 6||C = neg. U = neg||Dorsal pancreatic and splenic (1·3)||Hepatic (1)||Body/tail||Inferior body||Insulinoma|
| 7||M = neg||Distal splenic (17·5)||Gastroduodenal (1)||Tail||Tail||Insulinoma|
| 8||C = neg||Hepatic (1·9)d||Nil available||Head||Head||Insulinoma|
| 9||M = tail||Superior pancreaticoduodenal (1·8)||Hepatic (0·7)||Mid body||Between body/tail||Insulinoma|
|10||C = neg, U = neg||Distal splenic (3·8)||Gastroduodenal (1·2)||Tail||Between body/tail||Insulinoma|
|11||C = neg||Common hepatic (2·7)||Distal splenic (1·2)||Head||Nil||Normal pancreas|
|Group B: Calcium infusion 0·0065 mEq/kg|
|12||C = neg, M = tail||Prox splenic (62)||Gastroduodenal (1·1)||Body||Body||Insulinoma|
|13||C = neg, EU = neg P = head||Gastroduodenal (31)||Distal splenic (0·5)||Head||Head||Insulinoma|
|14||C = inferior tail M = body/tail||Prox splenic (12·7)||Gastroduodenal (1·1)||Body||Body||Insulinoma|
|15||C = tail||Mid splenic (4·4)||Gastroduodenal (1·5)||Tail||Tail||Insulinoma|
|16||C = neg, M = tail||Distal splenic (6·1)||Gastroduodenal (1·1)||Tail||Tail||Insulinoma|
|17||C = body, M = neg EU = neg||SMA (4)||Distal splenic (1·1)||Inferior head||Neck||Insulinoma|
|18||C = neg, EU = neg||Coeliac (2·4)||Gastroduodenal (1·7)||Uncertain||No surgery||No surgery|
|19||C = neg, M = ?tail 0 = neg||SMA (3·2)e||Distal splenic (2·4)||Head||Nil detected||Nesidioblastosis|
|20||C = neg, M = neg, EU = neg||Not identified||Not identified||Not identified||No surgery||No surgery|
|21||C = tail||Gastroduodenal (8·8)||Splenic (1·2)||Head||Not identifiedf||Normal pancreasf|
The results of the SACST test are shown in Table 2, together with the findings from preoperative pancreatic imaging. In Protocol A patients, a positive increment in serum insulin (>1·3) was seen in all 11 patients, and an insulinoma was found at the expected site in 10 individuals. Patient 11 had a positive insulin response (×2·7) in the common hepatic artery suggesting a lesion in the neck or head of the pancreas, and a lesser increment of 1·9 in the coeliac axis artery. A preoperative CT scan was nondiagnostic. No tumour was detected at surgery. A distal pancreatectomy was performed but the patient was not cured and histology showed normal islet tissue. The patient has subsequently been managed without specific therapy and has had only occasional further documented hypoglycaemia. Six patients in the Protocol B group had an insulinoma at the site predicted from the SACST study. Patient 19 demonstrated a significant insulin increment from infusion into the superior meseteric artery and was thought to most likely have a lesion near the head of the pancreas. No tumour was found at surgery, but a distal pancreatectomy showed nesidioblastosis. In retrospect, the distal splenic artery territory also showed a significant increment of 2·4, and the positive response in two widely separate arterial regions was consistent with diffuse islet cell disease. She has subsequently been managed with medical therapy. Patient 18 had a SACST result which was uncertain (positive in the coeliac axis, supplying multiple pancreatic regions), and elected not to proceed to surgery and has been treated medically. Patient 20 did not show a significant increment with calcium infusion in any arterial territory with SACST and did not localize using other imaging and has also been treated medically. The SACST in patient 21 predicted a tumour in the pancreatic body, whereas CT scan indicated a lesion ‘adjacent to’ the pancreatic tail which was not detected at surgery and which persists on CT scan and may be unrelated (? accessory spleen). This patient has unfortunately defaulted from further follow-up.
Analysis of the SACST results with Protocol A indicates a sensitivity of 91% for tumour localization and may indeed be 100% since the predicted lesion in the pancreatic neck in patient 11 has not been excluded histologically. Patient 11 has however been considered a false-positive result. With Protocol B, the correct prediction from SACST was six of eight patients coming to surgery (75%). The 25% false-positive localization may be less than stated, since the tumour localization in patient 21 remains uncertain. As mentioned above, patient 19 was probably misclassified in the SACST analysis. Patients 18 and 20 did not undergo surgery so the localizing ability of the SACST procedure is uncertain; a tumour was not localized in either patient. If these are considered, in the worst case scenario, as incorrect localization, then the positive prediction rate falls to 6/10 (60%) with Protocol B.
- Top of page
- Conflict of interest
This study compares the use of two different concentrations of calcium in the performance of the SACST procedure for localization of insulinoma in patients with biochemically proven hypoglycaemia and endogenous insulin excess. The results indicate that calcium concentrations lower than in the original published procedure are effective in provoking insulin release from insulinomas and can localize tumours within the pancreas. Protocol A, using one tenth of the calcium concentration described by Doppman et al. correctly identified the site of an insulinoma in 10 of 11 patients, using subsequent surgical localization as the gold standard. There was one false-positive result (patient 11) in whom an arterial territory showing maximum insulin release was thought to indicate a tumour in the head/neck region of the pancreas, but no tumour was identified at surgery. Normal pancreatic tissue was found histologically in the tail of the pancreas. It remains possible that this patient has a small tumour in the head of the pancreas which was not apparent at surgery, but the patient has been classified as a false positive.
Protocol B, using one quarter of the original calcium concentration suggested by Doppman et al. and utilized by O'Shea et al. correctly identified the site of an insulinoma in six of eight patients coming to surgery. Patient 19 was considered a false-positive result, with the SACST indicating a tumour in the head of the pancreas, but in retrospect, there was a significant although lesser insulin release with calcium injection in another arterial territory at the opposite end of the pancreas, which should have alerted us to the possibility of nesidioblastosis, which was the eventual diagnosis. The final diagnosis in patient 21 remains uncertain, with a mass adjacent to the pancreatic tail seen on CT scan but not identified or removed at surgery, whereas the SACST identified a likely tumour in the body of the pancreas. This patient has defaulted from follow-up so the final diagnosis remains uncertain, but for the purposes of this study, the patient has been classified as a false-positive result.
Assessing patients 11, 19 and 21 as false-positive localization by SACST indicates a correct overall positive detection rate of 16/19 patients (84%). This compares with correct localization rates in other reports of 63% (Druce et al.), 93% (Placzkowski et al.) and 84% (Guettier et al). A false-positive localization rate of 3/19 (16%) compares with rates of 4% in the NIH series, and 15% in the report from St Bartholomew's Hospital.
The rates of correct localization in the present study appear to be superior to those obtained with other modalities of localization using noninvasive scanning. In those with tumours localized by surgery, CT scan (in 15 patients) correctly identified the tumour site in just two subjects, with one false positive and 12 false negatives. The corresponding figures for transabdominal ultrasound were one true positive, two false negatives and two false positives, and for MRI scan three true positives, two false positives and one false negative. Photon emission tomography and octreoscan were performed in one subject each with one false positive (octreoscan) and one true positive (PET). Localization by conventional imaging agreed with SACST findings in seven patients. However, SACST provided correct localization in seven subjects in whom conventional imaging was either negative (n = 5) or not done (n = 2), and in two cases where conventional imaging gave incorrect results. In the other major series,[3, 12, 13] CT scan had a sensitivity of 32–57%, MRI scan 25–66% and transabdominal ultrasound 14–61% for correct tumour localization. Endoscopic ultrasound was only obtained in two patients in the current series, both with false-negative results. However, in experienced hands this technique has been reported to provide localization in up to 90% of patients with tumours in the head of the pancreas. The results of intraoperative ultrasound were only recorded in a minority of patients in the current series but can be helpful in those undergoing surgery.
No hypoglycaemic episodes were recorded with either calcium dose. This side effect is extremely rare in reports of the use of SACST, mainly because intravenous glucose and frequent glucose monitoring is used to maintain euglycaemia. However, severe hypoglycaemia was noted in one patient by O'Shea et al. using the standard calcium dose described by Doppman, which led to them adopting a lower calcium concentration of 0·0065 mEq/kg as used in Protocol B in the present study. The severity of flushing and tingling in the current report appeared less using the lowest calcium dose (Protocol A).
Use of lower doses of calcium than in the standard procedure required evaluation of what constitutes a normal response from uninvolved pancreatic tissue compared with a positive (exaggerated) response from a region of the pancreas containing an insulinoma. The Doppman protocol, adopted in other reports,[3, 12, 13] defines a twofold increase in insulin as indicating a positive response. The reliability of this criterion does not seem to have been rigorously evaluated but may have been drawn from the results of studies using transhepatic portal venous sampling. In the present study, the mean increment in insulin release from the insulinoma tissue using protocol A was a fourfold increase (range 1·3–17·5) and for the higher dosage (protocol B) was an 11-fold increment (range 2·4-62). Using a twofold increase in hepatic venous insulin concentration as indicating a positive response would not change the results in group B but would have altered the results in group A by reclassifying three individuals (patients 7, 9 and 10) as having negative tests. By assessing increments from injection into arteries supplying noninvolved regions of the pancreas remote from the site of maximum insulin release, we were able to define an appropriate cut-off for significant insulin release. In protocol A, the appropriate detection of insulinomas at surgery using this criterion (a 1·3-fold increase or greater being abnormal) in patients six, eight and nine supports this definition. However, if a twofold increase in insulin concentration is adopted as defining a positive response for all patients, the sensitivity of the procedure in group A would fall to 7 of 11 (64%) with three false negatives and one false-positive result.
The present study was performed in a limited number of patients, so the results should be interpreted with caution. Nonetheless, the study indicates that successful localization of insulinomas can be achieved by SACST using lower concentrations of calcium than previously described, and this reduction in calcium dose per artery may be associated with a lower rate of side effects. Although both calcium doses provided satisfactory localization rates, protocol B is probably preferable for routine use given published experience with this calcium dosage. However, correct localization of insulinomas remains a challenge, with a low rate of correct localization using conventional noninvasive scans and a small but significant rate of false-positive localization with SACST in this and other reports. Overall, SACST provides the best current means for localization of small insulinomas.