The present study describes the use of general clinical characteristics and biochemical indicators to establish a simple and feasible model for the diagnosis of insulinoma. This model was highly sensitive and accurate and may provide a simple and rapid method to diagnose insulinoma in patients with hypoglycemia and can be used in outpatients to diagnose insulinoma patients, which is helpful for diagnosis of more outpatients with insulinoma.
Insulinoma can cause severe metabolic disorders. Long-term hypoglycemia can cause irreversible damage to nerve tissue, and even endanger the life of the patient in severe cases . Therefore, early diagnosis and treatment are required. Insulinomas can be diagnosed qualitatively or based on localized factors . Qualitative diagnosis mainly relies on clinical manifestations and laboratory tests. At present, the 72-hour fasting test is the standard method for the qualitative diagnosis of insulinoma [6, 17–19]. An insulin-to-glucose ratio > 0.3 at the onset of hypoglycemia may provide a basis for diagnosing insulinoma. Many patients refuse to take the 72-hour fasting test, due to its being a painful experience and its accompanying risk of hypoglycemia. A retrospective analysis of the results in 69 patients with confirmed insulinoma showed that 20 (29.0%) had negative results on 72-hour fasting tests , indicating that negative results on 72-hour fasting tests cannot completely rule out a diagnosis of insulinoma . Insulinomas can also be diagnosed by imaging modalities, including ultrasound, CT, MRI, and endoscopic ultrasound (EUS), although the positivity rates of these noninvasive examinations are not high, with preoperative CT and MRI having sensitivity rates of 72% and 75%, respectively . EUS, another minimally invasive method, was found to have a sensitivity of 94% , although its sensitivity was largely dependent on the operator’s technique and experience. The sensitivities of EUS in detecting insulinomas in the head and body of the pancreas were high, at 95% and 98%, respectively , whereas its sensitivity in detecting lesions in the pancreatic tail was much lower, ranging from 37%-50% [25, 26]. Of the 37 patients pathologically diagnosed with insulinoma in the present study, 32 (86.5%) underwent EUS, with all 32 having space-occupying lesions, of minimum diameter 5.6 mm, in the tail of pancreas. The insulinoma patients included in our model included four false-negative cases, all of whom had tumors < 1cm in size. These findings suggested that application of this model yielded errors in patients with atypical clinical symptoms due to the small size of these tumors. Other methods, such as PET-CT and GLP-1 receptor imaging, have been shown superior to MRI and CT in diagnosing insulinomas. However, their clinical application is limited due to the high costs of examinations.
Weight gain is a significant manifestation of insulinoma. Due to the frequent occurrence of hypoglycemia symptoms, patients can relieve symptoms such as palpitation, tremor, and dizziness through eating . According to a retrospective study, 72% of patients with insulinoma have gained weight . In our study, the BMI of insulinoma patients was in the overweight range, while the BMI of the control group was within the normal range. Among the biochemical indicators, the HbA1c of patients in the insulinoma group was significantly lower than that in the control group, which suggests that the plasma glucose of patients with insulinoma has been low for a long time. Combining plasma glucose, insulin and C-peptide, insulinoma patients have higher fasting insulin and fasting C-peptide than the control group, indicating that insulinoma patients are more likely to have fasting hypoglycemia symptoms. After taking 75g anhydrous glucose powder, the indicators of insulinoma group, including 1-hour plasma glucose and 2-hours plasma glucose, are lower than the control group. Considering that insulinoma patients release a large amount of insulin, the improvement of plasma glucose level after taking anhydrous glucose powder is still relatively slow. The effect of insulinoma on patients' plasma glucose is a relatively long-lasting process, which is extremely harmful to the human body.
Insulinomas are relatively rare and differ in clinical symptoms, making this condition easy to miss and misdiagnose [28–30]. A clear diagnosis of insulinoma is a prerequisite for standardized treatment. Fajans’ and Turner’s indices are often used to evaluate the role of insulin release from pancreatic beta cells in regulating plasma glucose. Compared with indices, our model was more accurate and more reliable. A similar model of diagnosing insulinoma includes fasting insulin, fasting C-peptide, 1-hour C-peptide, and 2-hour insulin concentrations . The AUC of this model was 0.97, with a sensitivity of 86.5% and a specificity of 95.2%. That model could not be validated because the sample size was too small. Incorporation of our data into this model yielded an AUC of 0.957, a: sensitivity of 70.3%; and a specificity of 65.9%. The sensitivity and specificity of this model were lower than those of a model based on fasting plasma glucose and HbA1c concentrations in 82 patients with insulinoma and 100 normal controls . Because this control group consisted of normal persons, the results may not be as accurate. Similarly, this model could not be validated because of the small sample size. Compared with these earlier models, which used control groups consisting of normal healthy individuals, our model used a control group consisting of patients with hypoglycemia not caused by insulinoma, making our model more reliable. In addition, our model combined BMI with glucose-related indicators and used both single factor and multi-factor analysis to obtain the optimal formula. The data included in our model consisted of routine screening parameters for patients with hypoglycemia, providing this model high clinical feasibility and easy implementation.
This study had several limitations, such as the exclusion of patients with incomplete data, thus reducing the sample size. Moreover, this study was a single-center retrospective study, which may have introduced selection bias. Moreover, the small sample size prevented verification of the model. Additional studies, in larger numbers of patients, are needed to verify the accuracy of this model in diagnosing insulinomas. We will continue to collect general information of insulinoma patients and laboratory test results to further verify the model.