|
|
| |||||||||||||||||||||||||||||||
Background A lack of appropriate autonomic warning symptoms before the development of neuroglycopenia occurs frequently in patients with diabetes mellitus. The pathogenesis of this phenomenon is unclear, but it is associated with intensive insulin therapy, prolonged duration of diabetes, frequent episodes of hypoglycemia, and impaired glucose counterregulation. Recently, it has been proposed that repeated episodes of hypoglycemia may themselves induce the phenomenon.
Methods To test this hypothesis and to determine whether the phenomenon is reversible, we assessed autonomic and neuroglycopenic symptoms, counterregulatory hormonal responses, and cognitive function during stepped hypoglycemic-clamp studies in 6 patients with insulinomas before and approximately six months after curative surgery and in 14 normal subjects matched for age, weight, and sex.
Results Before surgery, the patients with insulinomas had lower scores than the normal subjects for autonomic symptoms (mean [±SD], 3.5 ±0.8 vs. 9.6 ±4.5) and neuroglycopenic symptoms (2.8 ±1.5 vs. 8.9 ±5.3). The patients also had impaired counterregulatory hormonal responses (their plasma epinephrine, norepinephrine, glucagon, growth hormone, and cortisol responses before surgery were 187 ±227 pg per milliliter [1.03 ±1.25 nmol per liter], 223 ±85 pg per milliliter [1.32 ±0.50 nmol per liter], 86 ±21 ng per liter, 7.4 ±5.2 µg per liter, and 12.1 ±1.5 µg per deciliter [334 ±41 nmol per liter], respectively, as compared with 842 ±439 pg per milliliter [4.63 ±2.41 nmol per liter], 519 ±150 pg per milliliter [3.07 ±0.89 nmol per liter], 201 ±58 ng per liter, 25.3 ±13.7 µg per liter, and 26.3 ±1.2 µg per deciliter [726 ±33 nmol per liter] in the normal subjects) and less deterioration in cognitive function than the normal subjects during hypoglycemia (sum of z scores for seven tests of cognitive function, 1.7 ±1.9 vs. 8.9 ±3.5) (P<0.02 for all comparisons). Surgical cure reversed all these abnormalities (P not significant for all comparisons with the normal subjects).
Conclusions Hypoglycemia itself can induce unawareness of the autonomic and neuroglycopenic symptoms of hypoglycemia and decrease the counterregulatory hormonal responses to hypoglycemia.
Because a decreased awareness of impending hypoglycemia can prevent patients from taking timely protective steps, this phenomenon can lead to more frequent and more severe episodes of hypoglycemia. It has been proposed that hypoglycemia itself may induce this unawareness of hypoglycemia1,10,11. The basis for this hypothesis is the observation that in both normal subjects and patients with type I diabetes, experimentally induced hypoglycemia reduces the subsequent awareness of hypoglycemia and counterregulatory hormonal responses10,11,12. A major problem with this hypothesis is that diabetic patients with unawareness of hypoglycemia have been reported to have no alteration in the development of neuroglycopenia,5 whereas experimentally induced hypoglycemia reduces both autonomic and neuroglycopenic symptoms10,13. This discrepancy raises doubt about whether experimentally induced hypoglycemia truly mimics the episodes of hypoglycemia experienced by diabetic patients in everyday life.
A clinical situation that may mimic the experience of diabetic patients who have frequent hypoglycemia is that of patients with insulinomas. Several case reports14,15,16,17,18,19,20,21 suggest that such patients have reduced awareness of hypoglycemia and reduced counterregulatory hormonal responses similar to those that occur in diabetic patients with unawareness of hypoglycemia. To test the hypothesis that hypoglycemia itself may induce the unawareness of hypoglycemia and to determine whether the phenomenon is reversible, we assessed autonomic and neuroglycopenic symptoms, cognitive function, and glucose counterregulatory hormonal responses during stepped hypoglycemic-clamp studies in six patients with insulinomas before and after surgical cure and in normal subjects matched for age, weight, and sex.
Methods
We studied 6 patients with insulinomas and 14 normal subjects (Table 1). The patients had had symptoms suggestive of hypoglycemia for 9 to 36 months (mean [±SD], 18 ±4). At the time of diagnosis, all the patients had at least one episode of hypoglycemia each day. Each patient was studied twice, once before surgery and once after. The interval between the two studies was 6.5 ±1.1 months, and the interval between surgery and the second study was 6.3 ±1.0 months. The results in 8 of the 14 normal subjects were included in an earlier report22. The protocol was approved by the institutional review committee, and all the study subjects gave written informed consent.
|
A semiquantitative symptoms questionnaire22,23 was administered every 15 minutes during the insulin infusions. The subjects scored each of the following symptoms on a scale from 0 (none) to 5 (severe): dizziness, tingling, blurred vision, difficulty in thinking, faintness, anxiety, palpitations, hunger, sweating, irritability, and tremor. On the basis of the categorization we and others have used,22,24 we considered the first five symptoms neuroglycopenic and the other six autonomic. The sum of each of these subcategories constituted the respective symptom score. In addition, the following standard cognitive tests were administered during each of the plateaus: trail-making part B, verbal fluency, digit vigilance, trail-making part A, verbal memory, and forward and backward digit span22. On the evening before a study, each subject repeatedly practiced each test. For the actual study, six alternative forms of each test were used. As previously described,22 the results of each cognitive test were transformed to unitless z scores (for each, the individual value minus the mean value, divided by the standard deviation of the group mean)25; the change from base line in the sum of all z scores was used for statistical analysis.
Analytical Methods
Plasma glucose was measured with a glucose analyzer (Yellow Springs Instruments, Yellow Springs, Ohio). Glycosylated hemoglobin was measured by high-performance liquid chromatography (Diamat, Bio-Rad, Richmond, Calif.; normal range, 4.3 to 6.1 percent). Plasma insulin, glucagon, growth hormone, epinephrine, norepinephrine, and cortisol were measured in duplicate as previously described20,22.
Statistical Analysis
Glycemic thresholds were calculated for the various measurements (e.g., the symptoms score or plasma epinephrine concentration) as previously described22,23. First, the results were adjusted to a base line of zero by subtracting the mean of three base-line values from all the later values for that study. For each sampling period, the 95 percent confidence interval of the group mean had been determined for each measurement in previously reported studies performed in euglycemic normal subjects22. The results of each measurement during the stepped hypoglycemic-clamp studies were then compared individually with the 95 percent confidence intervals for the corresponding measurements from the euglycemic-control study. In each subject, the plasma glucose concentration at which a given variable first consistently exceeded the 95 percent confidence interval was designated the glycemic threshold for that variable. The maximal responses for plasma hormone concentrations, symptom scores, and cognitive scores during the six-hour period were used to compare the magnitude of responses. The results are given as means ±SD unless otherwise indicated. We used the Mann-Whitney test25 to compare the results in the normal subjects with those in the patients with insulinomas and the Wilcoxon test25 to compare the results in the patients before and after surgery. Correlations were examined with least-squares linear regression analysis25.
Results
Clinical and Biochemical Characteristics of the Study Subjects
Before surgery, the patients with insulinomas had hypoglycemia while fasting, hyperinsulinemia, and lower glycosylated hemoglobin values than the normal subjects (Table 1). After surgery the fasting plasma glucose and insulin concentrations and glycosylated hemoglobin values of the patients were no longer significantly different from those of the normal subjects.
Glucose Counterregulation
Plasma glucose concentrations during the stepped hypoglycemic-clamp studies were similar in all three sampling periods (Figure 1). Before surgery the responses of all counterregulatory hormones (glucagon, epinephrine, norepinephrine, growth hormone, and cortisol) began at lower plasma glucose concentrations in the patients with insulinomas than in the normal subjects (P<0.02 for all comparisons) and were smaller (P<0.02 for all comparisons) (Figure 2 and Table 2). The responses in the patients were 187 ±227 pg per milliliter (1.03 ±1.25 nmol per liter) for epinephrine, 223 ±85 pg per milliliter (1.32 ±0.50 nmol per liter) for norepinephrine, 86 ±21 ng per liter for glucagon, 7.4 ±5.2 µg per liter for growth hormone, and 12.1 ±1.5 µg per deciliter (334 ±41 nmol per liter) for cortisol, as compared with 842 ±439 pg per milliliter (4.63 ±2.41 nmol per liter), 519 ±150 pg per milliliter (3.07 ±0.89 nmol per liter), 201 ±58 ng per liter, 25.3 ±13.7 µg per liter, and 26.3 ±1.2 µg per deciliter (726 ±33 nmol per liter), respectively, in the normal subjects. After surgery, the thresholds and magnitude of all responses improved in the patients and were no longer significantly different from those of the normal subjects.
|
|
|
Before surgery, the patients with insulinomas were often asymptomatic when their plasma glucose concentrations were below 35 mg per deciliter (1.9 mmol per liter). Indeed, one woman with an insulinoma was writing a letter and conversing coherently when her plasma glucose concentration was 29 mg per deciliter (1.6 mmol per liter). This reduced awareness of hypoglycemia with preserved cognitive function was quantitated during the stepped hypoglycemic-clamp studies (Table 2 and Figure 3). Before surgery the threshold for autonomic symptoms occurred at a lower mean plasma glucose concentration in the patients with insulinomas than in the normal subjects (45 ±4 vs. 60 ±6 mg per deciliter [2.5 ±0.2 vs. 3.3 ±0.2 mmol per liter], P<0.001), and the magnitude of the symptoms was less in the patients (score, 3.5 ±0.8 vs. 9.6 ±4.5; P = 0.001). After surgery, both the threshold and the magnitude returned to normal. Although the patients' threshold for the onset of neuroglycopenic symptoms before surgery was not significantly different from that of the normal subjects (50 ±9 vs. 52 ±8 mg per deciliter [2.8 ±0.5 vs. 2.9 ±0.4 mmol per liter], P = 0.65), the magnitude of their symptoms was significantly less (score, 2.8 ±1.5 vs. 8.9 ±5.3; P = 0.001) (Figure 3). Moreover, the plasma glucose concentration at which neuroglycopenic symptoms began increased in all the patients after surgery (mean concentration before surgery, 50 ±9 mg per deciliter [2.8 ±0.5 mmol per liter]; after surgery, 54 ±10 mg per deciliter [3.0 ±0.5 mmol per liter]; range of increase, 1 to 10 mg per deciliter [0.06 to 0.6 mmol per liter]; P = 0.027) and the magnitude reverted to normal (score, 8.3 ±5.4; P = 0.027).
|
In contrast to the normal subjects, in whom cognitive function deteriorated during the stepped hypoglycemic-clamp study (z score at last glycemic plateau, 8.9 ±3.5; P<0.001 for the comparison with base line), there was no significant change in cognitive function in the patients with insulinomas before surgery (z score at last glycemic plateau, 1.7 ±1.9; P = 0.075 for the comparison with base line and P = 0.002 for the comparison with the normal subjects) (Figure 4). After surgery, the patients with insulinomas had a degree of cognitive deterioration (z score at the last glycemic plateau, 9.1 ±5.1; P<0.001 for the comparison with base line) that was not significantly different from that of the normal subjects (P = 0.62). Although it was not possible to determine a threshold for cognitive deterioration in the patients with insulinomas before surgery, they were assigned a threshold equal to their lowest plasma glucose concentration during the clamp study in order to permit statistical comparison. This threshold (39 ±2 mg per deciliter [2.2 ±0.1 mmol per liter]) was significantly different from that of the normal subjects (46 ±3 mg per deciliter [2.6 ±0.2 mmol per liter], P = 0.003) and from that of the patients after surgery (49 ±4 mg per deciliter [2.7 ±0.2 mmol per liter], P = 0.027). When all the study subjects were considered together, the thresholds for cognitive deterioration were significantly correlated with the glycosylated hemoglobin values (r = 0.45, P<0.05). In addition, the glycosylated hemoglobin values were significantly correlated with the thresholds for autonomic symptoms (r = 0.47, P<0.05) and for initiation of the plasma epinephrine (r = 0.57, P<0.01) and norepinephrine (r = 0.44, P<0.05) responses, but not for the onset of neuroglycopenic symptoms.
|
We found that untreated patients with insulinomas had reduced autonomic and neuroglycopenic symptoms, impaired counterregulatory hormonal responses, and diminished deterioration of cognitive function during hypoglycemia. All these abnormalities were reversed after surgical removal of the tumors. Our results confirm reports of previous cases16,17,18,20,21 in which some of the same tests were performed before and after surgery.
These results have implications regarding the pathogenesis of unawareness of hypoglycemia in patients with diabetes mellitus. It has been postulated1,10,11 that episodes of hypoglycemia may cause the phenomenon in patients with diabetes because experimentally induced hypoglycemia reduces awareness of hypoglycemia (manifested as both autonomic and neuroglycopenic symptoms) and impairs counterregulatory hormonal responses during hypoglycemia.
Some investigators,5 however, have reported that diabetic patients with unawareness of hypoglycemia have a selective reduction in autonomic warning symptoms, whereas others26 have reported changes in both autonomic and neuroglycopenic symptoms. The results of this study of patients with insulinomas indicate that repeated spontaneous episodes of hypoglycemia, independently of diabetes mellitus, reduce both autonomic and neuroglycopenic symptoms and cause cognitive dysfunction during induced hypoglycemia -- findings that support the view that hypoglycemia itself induces unawareness of hypoglycemia in diabetic patients. A possible cause of the discrepancies with respect to neuroglycopenic symptoms may be related to their subjective nature. For example, although we found no significant correlation between glycosylated hemoglobin values and the threshold for neuroglycopenic symptoms, we did find a significant correlation between those values and the threshold for objectively determined cognitive dysfunction.
It may seem paradoxical that diabetic patients with unawareness of hypoglycemia have an increased frequency of severe hypoglycemia when their neuroglycopenia develops at lower plasma glucose concentrations. As this study illustrates, however, the interval (in terms of plasma glucose concentration) between the onset of autonomic warning symptoms and the development of neuroglycopenia is reduced in patients with unawareness of hypoglycemia. Thus, the time that a patient with unawareness of hypoglycemia has to respond to warning signals and take protective action is diminished.
Finally, this and other studies16,17,18,20,21 in which patients with insulinomas were investigated before and after surgery have implications regarding possible treatment of the unawareness of hypoglycemia in diabetic patients. Since the phenomenon was completely reversed in patients with insulinomas by the elimination of their hypoglycemia, it may also be reversible in diabetic patients if their episodes of hypoglycemia can be prevented. Indeed, adjustments in diet and insulin treatment that reduced episodes of hypoglycemia improved awareness of hypoglycemia in five patients with type I diabetes27. Whether these results are generally applicable to patients with type I diabetes and whether they can be achieved while satisfactory glycemic control is maintained remain to be determined.
Supported in part by grants from the National Institutes of Health (RR00056, RR00036, DK29411, DK2085, DK39629, DK44235, and DK20479); the Consiglio Nazionale della Richerche (CC.N.R. 90 20412CT04 and CNR 910035PF40); and the Foundation "De Drie Lichten," the Netherlands. Dr. Veneman received a mentor-based research fellowship from the American Diabetes Association, and Dr. Perriello was a Fogarty Foundation Fellow.
We are indebted to the staff of the General Clinical Research Center, Carol Korbanic, Daniel Kirsch, Dawn Purdy, and Suresh Shah, for their excellent technical help, and to Laura Brinker for superb editorial assistance.
Source Information
From the Second Department of Internal Medicine, Propaedeutic, Athens University, Athens, Greece (A.M., D.P., S.R.); the Istituto di Medicina Interna e Scienze Endocrine e Metaboliche, University of Perugia, Perugia, Italy (C.F., G.P., S.C., A.R., P.B., G.B.); the Whittier Institute for Diabetes and Endocrinology, La Jolla, Calif. (T.V., J.G.); and the Department of Medicine, Washington University School of Medicine, St. Louis (P.C.).
Address reprint requests to Dr. Gerich at the Whittier Institute for Diabetes and Endocrinology, 9894 Genesee Ave., La Jolla, CA 92037.
References
| |||||||||||||||||||||||||||||||
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
Previous
