Glucose Tolerance and Serum Insulin in Identical Twins of Diabetics

Glucose tolerance with serum insulin assay has been carried out in 24 apparently unaffected identical twins of diabetics. Glucose values were significantly higher and insulin values significantly lower than in control subjects, but there was considerable individual variation and in half the cases glucose tolerance was normal. There was no case of increased insulin response. In these twins glucose tolerance and insulin response were not correlated with weight, family history of diabetes, age at diagnosis of diabetes in the affected twin, time since that diagnosis, nor with age at testing the unaffected twin. On retesting 12 of the twins after two years no significant differences were found from the first testing.We conclude that not all these twins are likely to develop diabetes, that the assumption that identical twins of diabetics are necessarily “prediabetic” is probably erroneous, and that factors other than genetic ones are important in the aetiology of diabetes.     

Control of Basal Insulin Secretion,with Special Reference to the Diagnosis of Insulinomas

Estimation of plasma glucose and immunoreactive insulin concentrations in normal subjects after an overnight fast showed that subjects with high basal plasma glucose levels tended to have high plasma insulin concentrations. A similar correlation between glucose and insulin levels was seen in patients with obesity and various endocrine disorders. The suppression of plasma insulin levels associated with hypoglycaemia was used to derive an “amended insulin-glucose ratio,” which appeared to be a good discriminant for the diagnosis of insulinomas. In normal subjects the ratio was less than 30 μU insulin/mg glucose, in obese subjects less than 50 μU/mg, and most of the patients with insulinomas had values over 200 μU/mg.     

Myocardial Infarction and Carbohydrate Metabolism Relating to Diabetes Mellitus

Fifteen non-obese males with acute myocardial infarction and no diabetic history were evaluated for diabetes. During infarction, results of oral glucose tolerance tests were “diabetic” or “probably diabetic” in 10 of the 15 patients (67 percent). The plasma immuno-reactive insulin response in 12 patients (80 percent) was of a pattern observed in patients with maturity-onset diabetes. Six months after infarction, follow-up glucose tolerance tests in 12 surviving patients were diabetic or probably diabetic in three cases (25 percent). In seven of twelve patients (58 percent) had delay in the peaking of the plasma insulin response to an oral glucose tolerance test, a phenomenon that is observed in patients with maturity-onset diabetes.Glucose tolerance tests were abnormal in one of fourteen control subjects (7 percent). There was a delayed plasma insulin response to an oral glucose test in two of fourteen controls (14 percent).Patients with myocardial infarction have an increased incidence of diabetes mellitus.     

Ketonaemia in Uncontrolled Diabetes Mellitus

Blood ketone bodies, serum insulin levels, and plasma free fatty acids were examined in a series of patients with “non-ketotic diabetic coma” and compared with the findings in ketoacidotic subjects. Serum insulin levels in six “non ketotic” patients ranged between 1 and 25 μu./ml. and were not significantly different from levels reported in patients with ketoacidosis. In addition, plasma free fatty acids were shown to be unrelated to the degree of ketonaemia. The investigation shows that neither the levels of serum insulin nor those of free fatty acids can explain the absence of hyperketonaemia in some cases.     

Influence of the Time of Day and Fasting Duration on Glucose Level following a 1-Hour, 50-Gram Glucose Challenge Test in Pregnant Women

Background

Previous studies have shown that the time of day (TD) of glucose measurement and the fasting duration (FD) influence the glucose levels in adults. Few studies have examined the effects of the TD and FD on the glucose level following a 1-hour, 50-gram glucose challenge test (GCT) in pregnant women in screening for or diagnosing gestational diabetes mellitus (GDM). The objective of this study was to investigate the influence of the TD (morning, afternoon, night) and the FD (the time of the last food ingestion as follows: ≤1 hour, 1–2 hours, and >2 hours) by examining their combined effects on the glucose levels following a 50-gram GCT in pregnant women.

Methods and Results

We analyzed the data of 1,454 non-diabetic pregnant Taiwanese women in a prospective study. Multiple linear regression and multiple logistic regression were used to estimate the relationships between the 9 TD-FD groups and the continuous and binary glucose levels (cut-off at 140 mg/dL) following a 50-gram GCT, after adjusting for maternal age, nulliparity, pre-pregnancy body mass index, and weight gain. Different TD and FD groups were associated with variable glucose responses to the 50-gram GCT, some of which were significant. The estimate coefficients (β) of the TD-FD groups “night, ≤1 hr” and “night, 1–2 hr” revealed significantly lower glucose concentrations [β (95% confidence interval [CI]): −6.46 (−12.53, −0.38) and −6.85 (−12.50, −1.20)] compared with the “morning, >2 hr” group. The TD-FD groups “afternoon, ≤1 hr” and “afternoon, 1–2 hr” showed significantly lower odds ratios (OR) of a positive GCT; the adjusted ORs (95% CI) were 0.54 (0.31–0.95) and 0.58 (0.35–0.96), respectively.

Conclusions

Our findings demonstrate the importance of standardizing the TD and FD for the 1-hour, 50-gram GCT. In screening for and diagnosing GDM, the TD and FD are modifiable factors that should be considered in clinical practice and epidemiological studies.     

Computational Modeling of Glucose Transport in Pancreatic β-Cells Identifies Metabolic Thresholds and Therapeutic Targets in Diabetes

Pancreatic β-cell dysfunction is a diagnostic criterion of Type 2 diabetes and includes defects in glucose transport and insulin secretion. In healthy individuals, β-cells maintain plasma glucose concentrations within a narrow range in concert with insulin action among multiple tissues. Postprandial elevations in blood glucose facilitate glucose uptake into β-cells by diffusion through glucose transporters residing at the plasma membrane. Glucose transport is essential for glycolysis and glucose-stimulated insulin secretion. In human Type 2 diabetes and in the mouse model of obesity-associated diabetes, a marked deficiency of β-cell glucose transporters and glucose uptake occurs with the loss of glucose-stimulated insulin secretion. Recent studies have shown that the preservation of glucose transport in β-cells maintains normal insulin secretion and blocks the development of obesity-associated diabetes. To further elucidate the underlying mechanisms, we have constructed a computational model of human β-cell glucose transport in health and in Type 2 diabetes, and present a systems analysis based on experimental results from human and animal studies. Our findings identify a metabolic threshold or “tipping point” whereby diminished glucose transport across the plasma membrane of β-cells limits intracellular glucose-6-phosphate production by glucokinase. This metabolic threshold is crossed in Type 2 diabetes and results in β-cell dysfunction including the loss of glucose stimulated insulin secretion. Our model further discriminates among molecular control points in this pathway wherein maximal therapeutic intervention is achieved.     

The “therapeutic range” of the one-stage prothrombin time in the control of anticoagulant therapy: The effect of different thromboplastin preparations

Commercially available thromboplastin reagents and two human brain preparations have been compared using the one-stage prothrombin time and plasma samples from patients receiving long-term oral anticoagulant therapy. Considerable variation is noted between various thromboplastins using the same plasma sample. The commercially available thromboplastins give shorter prothrombin times than do human brain preparations. With the latter, the “therapeutic range” is represented by a prothrombin time of about 1.8 to 3.0 times the normal control value, whereas with commercial preparations the “therapeutic range” is about 1.25 to 1.75 times normal. The implications of these observations are discussed; the desirability of standardization of the one-stage prothrombin time is emphasized.     

Circulating Docosahexaenoic Acid Levels Are Associated with Fetal Insulin Sensitivity

Background

Arachidonic acid (AA; C20∶4 n-6) and docosahexaenoic acid (DHA; C22∶6 n-3) are important long-chain polyunsaturated fatty acids (LC-PUFA) in maintaining pancreatic beta-cell structure and function. Newborns of gestational diabetic mothers are more susceptible to the development of type 2 diabetes in adulthood. It is not known whether low circulating AA or DHA is involved in perinatally “programming” this susceptibility. This study aimed to assess whether circulating concentrations of AA, DHA and other fatty acids are associated with fetal insulin sensitivity or beta-cell function, and whether low circulating concentrations of AA or DHA are involved in compromised fetal insulin sensitivity in gestational diabetic pregnancies.

Methods and Principal Findings

In a prospective singleton pregnancy cohort, maternal (32-35 weeks gestation) and cord plasma fatty acids were assessed in relation to surrogate indicators of fetal insulin sensitivity (cord plasma glucose-to-insulin ratio, proinsulin concentration) and beta-cell function (proinsulin-to-insulin ratio) in 108 mother-newborn pairs. Cord plasma DHA levels (in percentage of total fatty acids) were lower comparing newborns of gestational diabetic (n = 24) vs. non-diabetic pregnancies (2.9% vs. 3.5%, P = 0.01). Adjusting for gestational age at blood sampling, lower cord plasma DHA levels were associated with lower fetal insulin sensitivity (lower glucose-to-insulin ratio, r = 0.20, P = 0.036; higher proinsulin concentration, r = −0.37, P <0.0001). The associations remained after adjustment for maternal and newborn characteristics. Cord plasma saturated fatty acids C18∶0 and C20∶0 were negatively correlated with fetal insulin sensitivity, but their levels were not different between gestational diabetic and non-diabetic pregnancies. Cord plasma AA levels were not correlated with fetal insulin sensitivity.

Conclusion

Low circulating DHA levels are associated with compromised fetal insulin sensitivity, and may be involved in perinatally “programming” the susceptibility to type 2 diabetes in the offspring of gestational diabetic mothers.     

Improved Metabolic Health Alters Host Metabolism in Parallel with Changes in Systemic Xeno-Metabolites of Gut Origin

Novel plasma metabolite patterns reflective of improved metabolic health (insulin sensitivity, fitness, reduced body weight) were identified before and after a 14–17 wk weight loss and exercise intervention in sedentary, obese insulin-resistant women. To control for potential confounding effects of diet- or microbiome-derived molecules on the systemic metabolome, sampling was during a tightly-controlled feeding test week paradigm. Pairwise and multivariate analysis revealed intervention- and insulin-sensitivity associated: (1) Changes in plasma xeno-metabolites (“non-self” metabolites of dietary or gut microbial origin) following an oral glucose tolerance test (e.g. higher post-OGTT propane-1,2,3-tricarboxylate [tricarballylic acid]) or in the overnight-fasted state (e.g., lower γ-tocopherol); (2) Increased indices of saturated very long chain fatty acid elongation capacity; (3) Increased post-OGTT α-ketoglutaric acid (α-KG), fasting α-KG inversely correlated with Matsuda index, and altered patterns of malate, pyruvate and glutamine hypothesized to stem from improved mitochondrial efficiency and more robust oxidation of glucose. The results support a working model in which improved metabolic health modifies host metabolism in parallel with altering systemic exposure to xeno-metabolites. This highlights that interpretations regarding the origins of peripheral blood or urinary “signatures” of insulin resistance and metabolic health must consider the potentially important contribution of gut-derived metabolites toward the host''s metabolome.     

Mechanism of action of insulin in diabetic patients: a dose-related effect on glucose production and utilisation

Six insulin-requiring diabetics were studied after insulin had been withheld for 24 hours. On three separate occasions each received a two-hour infusion of insulin at a low dose (2·6 U/h) and a high dose (10·6 U/h) and an infusion of saline as control. The rates of production and utilisation of glucose were measured isotopically. The rate of fall of plasma glucose concentration was faster on the high-dose infusion of insulin than on the low, whereas the fall in plasma free fatty acids, glycerol, and keton bodies was the same on both insulin infusions. The mechanism whereby the two rates of insulin administration lowered plasma glucose concentration differed: during the low-dose infusion the decrease in the glucose concentration was produced entirely by a fall of hepatic glucose output, whereas during the high-dose insulin infusion the glucose concentration fell because both the rate of glucose production fell and the rate of glucose utilisation rose. In all experiments there was a direct relation between a fall in serum potassium concentration and the fall in plasma glucose concentration irrespective of the mechanism that reduced the glucose concentration.These results indicate that in uncontrolled diabetics low-dose insulin infusions lower the blood glucose concentration entirely by reducing glucose production from the liver and that the effect of insulin on potassium transport is independent of its effect on glucose uptake.     

Red cell sorbitol and diabetic control

There is a disagreement about the value of red cell sorbitol as an indicator of diabetic control. We studied 7 insulin requiring diabetics, one for 10 days and 6 for 24 hours. In the 10-day study, blood samples were obtained every 4-6 hours. Sequential analysis of plasma glucose and red cell sorbitol levels showed that simultaneous levels had the best correlation r = 0.534, P less than .001. In the 24-hour studies, blood samples were obtained at 1 to 4-hour intervals. Three of the 6 patients showed the best correlation between simultaneous plasma glucose and red cell sorbitol, with r = 0.727 to 0.957 and P less than .001 to P less than .07. When plasma glucose values were compared to red cell sorbitol levels 1, 2, 4, 8 and 12 h later, correlation coefficients were not as good as the simultaneous comparison. We conclude that the measurement of red cell sorbitol correlates well with simultaneous plasma glucose in some but not all diabetic subjects and provides no additional information to the clinician.     

A Critical Point of Male Gonad Development: Neuroendocrine Correlates of Accelerated Testicular Growth in Rats during Early Life

Testis growth during early life is important for future male fertility and shows acceleration during the first months of life in humans. This acceleration coincides with the peak in gonadotropic hormones in the blood, while the role of hypothalamic factors remains vague. Using neonatal rats to assess this issue, we found that day 9 of life is likely critical for testis development in rats. Before this day, testicular growth was proportional to body weight gain, but after that the testes showed accelerated growth. Hypothalamic kisspeptin and its receptor mRNA levels begin to elevate 2 days later, at day 11. A significant increase in the mRNA levels for gonadotropin-releasing hormone (GnRH) receptors in the hypothalamus between days 5 and 7 was followed by a 3-fold decrease in GnRH mRNA levels in this brain region during the next 2 days. Starting from day 9, hypothalamic GnRH mRNA levels increased significantly and positively correlated with accelerated testicular growth. Triptorelin, an agonist of GnRH, at a dose that had no effect on testicular growth during “proportional” period, increased testis weights during the period of accelerated growth. The insensitivity of testicular growth to GnRH during “proportional” period was supported by inability of a 2.5-fold siRNA knockdown of GnRH expression in the hypothalamus of the 7-day-old animals to produce any effect on their testis weights. GnRH receptor blockade with cetrorelix was also without effect on testis weights during “proportional” period but the same doses of this GnRH antagonist significantly inhibited “accelerated” testicular growth. GnRH receptor mRNA levels in the pituitary as well as plasma LH concentrations were higher during “accelerated” period of testicular growth than during “proportional” period. In general, our data defined two distinct periods in rat testicular development that are primarily characterized by different responses to GnRH signaling.     

Photoperiodic Entrainment Patterns in the CO(2) Output of Lemna perpusilla 6746 and of Several Other Lemnaceae

The CO₂ output of Lemna perpusilla 6746 in “skeleton photoperiods” consisting of alternating 10½-hour and 13-hour dark periods separated by ¼-hour illuminations was recorded under stable high and low nitrate conditions. The phase relationship finally attained between light schedule and output is the same regardless of which dark period is given first, but entrainment is more rapid (as is flowering) with an initial 13-hour dark period. In all respects other than bistability—the assumption of two different stable phase relationships depending on the initial dark period—both flowering and the course of CO₂ output conform to Pittendrigh's model derived from Drosophila eclosion rhythms, confirming the view that an endogenous circadian rhythm, or biological clock, underlies the photoperiodic control of flowering in this plant. Experiments with rigorous temperature control show that earlier results with long light exposures were in part due to temperature changes; in consequence, it is clear that entrainment patterns with high nitrate differ even more from those in low nitrate than was previously evident, and not simply by the addition of a “nitrate peak.” Other Lemnaceae tested with a few simple light-dark schedules in both types of media show a variety of responses, with no obvious correlation to photoperiodic response type.     

Unrecognized ocular problems associated with port-wine stain of the face in children

The port-wine stain, a relatively rare subgroup of capillary hemangiomas, usually involves the face and in children it may be associated with glaucoma. A series of 50 patients with facial port-wine stains had ocular examination at The Hospital for Sick Children, Toronto. When the facial areas supplied by both the ophthalmic and maxillary divisions of the sensory branch of the trigeminal nerve were involved (26 patients) there was about a 45% chance of diagnosing glaucoma, either as “true” glaucoma (with visual loss and raised intraocular pressure) or as “glaucoma suspect” (without visual loss). Involvement of the area supplied by either division alone was not associated with glaucoma.     

Post-Weaning Diet Affects Faecal Microbial Composition but Not Selected Adipose Gene Expression in the Cat (Felis catus)

The effects of pre- (i.e., gestation and during lactation) and post-weaning diet on the composition of faecal bacterial communities and adipose expression of key genes in the glucose and insulin pathways were investigated in the cat. Queens were maintained on a moderate protein:fat:carbohydrate kibbled (“Diet A”; 35:20:28% DM; n  =  4) or high protein:fat:carbohydrate canned (“Diet B”; 45:37:2% DM; n = 3) diet throughout pregnancy and lactation. Offspring were weaned onto these diets in a nested design (n  =  5 per treatment). Faecal samples were collected at wk 8 and 17 of age. DNA was isolated from faeces and bacterial 16S rRNA gene amplicons were analysed by pyrosequencing. RNA was extracted from blood (wk 18) and adipose tissue and ovarian/testicular tissues (wk 24) and gene expression levels determined using RT-qPCR. Differences (P<0.05) in composition of faecal bacteria were observed between pregnant queens fed Diet A or B. However, pre-weaning diet had little effect on faecal bacterial composition in weaned kittens. In contrast, post-weaning diet altered bacterial population profiles in the kittens. Increased (P<0.05) abundance of Firmicutes (77% vs 52% of total reads) and Actinobacteria (0.8% vs 0.2% of total reads), and decreased (P<0.05) abundance of Fusobacteria (1.6% vs 18.4% of total reads) were observed for kittens fed the Diet A compared to those fed Diet B post-weaning. Feeding Diet B pre-weaning increased (P<0.05) the expression levels of INRS, LEPT, PAI-1 and tended to increase GLUT1, while the expression levels of IRS-1 in blood increased in kittens fed Diet A pre-weaning. Post-weaning diet had no effect on expression levels of target genes. Correlations between the expression levels of genes involved in glucose and insulin pathways and faecal Bacteriodetes and Firmicutes phyla were identified. The reasons for why post-weaning diet affects microbial populations and not gene expression levels are of interest.     

Growth Hormone Release Inhibiting Hormone in Acromegaly

Growth hormone release inhibiting hormone (GHRIH) was administered by constant infusion over 75 minutes to eight acromegalic patients at different doses. 100 to 1,000 μg were equally effective in reducing circulating growth hormone (GH) levels; 25 μg lowered GH levels in only five patients, and at this dose the extent of the fall was smaller than from doses of 100 μg or more. 10 μg was ineffective. Injection of single doses of 500 μg by intravenous, subcutaneous, and intramuscular routes caused only small and transient reductions in GH levels, though the effect was improved by injecting the hormone intramuscularly in 2 ml of 16% gelatin. Injection of a suspension of 4 mg GHRIH in 1 ml of arachis oil lowered growth hormone levels for between three and four hours.In four acromegalic patients an oral 50-g glucose tolerance test was performed during a continuous infusion of either saline or 1,000 μg GHRIH. The “paradoxical” rise in growth hormone seen in these patients during the saline infusion was suppressed by GHRIH. The blood glucose responses were, moreover, modified by GHRIH in that the peak was delayed and occurred at the end of the infusion in each case. A “normal” glucose tolerance curve was converted to a “diabetic” type of response in two patients. This effect could be accounted for by the inhibition of insulin secretion known to occur with large doses of GHRIH.We speculate that acromegaly may be primarily a hypothalmic disease due to deficiency of GHRIH resulting in excessive secretion of growth hormone from the pituitary and adenoma formation due to inappropriate and prolonged stimulation of the pituitary.     

A Novel Glycated Hemoglobin A1c-Lowering Traditional Chinese Medicinal Formula,Identified by Translational Medicine Study

Diabetes is a chronic metabolic disorder that has a significant impact on the health care system. The reduction of glycated hemoglobin A1c is highly associated with the improvements of glycemic control and diabetic complications. In this study, we identified a traditional Chinese medicinal formula with a HbA1c-lowering potential from clinical evidences. By surveying 9,973 diabetic patients enrolled in Taiwan Diabetic Care Management Program, we found that Chu-Yeh-Shih-Kao-Tang (CYSKT) significantly reduced HbA1c values in diabetic patients. CYSKT reduced the levels of HbA1c and fasting blood glucose, and stimulated the blood glucose clearance in type 2 diabetic mice. CYSKT affected the expressions of genes associated with insulin signaling pathway, increased the amount of phosphorylated insulin receptor in cells and tissues, and stimulated the translocation of glucose transporter 4. Moreover, CYSKT affected the expressions of genes related to diabetic complications, improved the levels of renal function indexes, and increased the survival rate of diabetic mice. In conclusion, this was a translational medicine study that applied a “bedside-to-bench” approach to identify a novel HbA1c-lowering formula. Our findings suggested that oral administration of CYSKT affected insulin signaling pathway, decreased HbA1c and blood glucose levels, and consequently reduced mortality rate in type 2 diabetic mice.     

Glucagon-Like Peptide-1 Induced Signaling and Insulin Secretion Do Not Drive Fuel and Energy Metabolism in Primary Rodent Pancreatic β-Cells

Background

Glucagon like peptide-1 (GLP-1) and its analogue exendin-4 (Ex-4) enhance glucose stimulated insulin secretion (GSIS) and activate various signaling pathways in pancreatic β-cells, in particular cAMP, Ca²⁺ and protein kinase-B (PKB/Akt). In many cells these signals activate intermediary metabolism. However, it is not clear whether the acute amplification of GSIS by GLP-1 involves in part metabolic alterations and the production of metabolic coupling factors.

Methodology/Prinicipal Findings

GLP-1 or Ex-4 at high glucose caused release (∼20%) of the total rat islet insulin content over 1 h. While both GLP-1 and Ex-4 markedly potentiated GSIS in isolated rat and mouse islets, neither had an effect on β-cell fuel and energy metabolism over a 5 min to 3 h time period. GLP-1 activated PKB without changing glucose usage and oxidation, fatty acid oxidation, lipolysis or esterification into various lipids in rat islets. Ex-4 caused a rise in [Ca²⁺]_i and cAMP but did not enhance energy utilization, as neither oxygen consumption nor mitochondrial ATP levels were altered.

Conclusions/Significance

The results indicate that GLP-1 barely affects β-cell intermediary metabolism and that metabolic signaling does not significantly contribute to GLP-1 potentiation of GSIS. The data also indicate that insulin secretion is a minor energy consuming process in the β-cell, and that the β-cell is different from most cell types in that its metabolic activation appears to be primarily governed by a “push” (fuel substrate driven) process, rather than a “pull” mechanism secondary to enhanced insulin release as well as to Ca²⁺, cAMP and PKB signaling.     

Is phospholipase A2 a “glucose sensor” responsible for the phasic pattern of insulin release?

The mechanisms involved in the characteristic, normal biphasic pattern of glucose-induced insulin release (which is grossly altered in type II diabetics) have not been definitely elucidated. However, the temporal pattern of arachidonic acid release induced by cellular phospholipases precisely mimics that of first-phase insulin release, both being characterized by a burst of release peaking near 2—3 minutes and followed by a “trough” or apparent refractory period most apparent at 5–10 minutes. The latter appears temporally related not only to decreased arachidonate release but also to stimulation of its re-esterification. Pancreatic islets contain a glucose-sensitive phospholipase A₂, and glucose has been shown to increase the accumulation of islet lipoxygenase-derived products which appear to be “third messengers” mediating insulin release. Blockade either of islet phospholiphase(s) or of islet lipoxygenase totally abrogates glucose-induced insulin release. The hypothesis is therefore proposed that phospholiphase A₂ could be one beta cell “glucose sensor”, and that the released arachidonate is coupled to an islet lipoxygenase. Labile oxygenated metabolites (lipid peroxides and epoxides) transduce the glucose signal into insulin release. The available data (albeit incomplete) are compatible with the formulation that the biphasic pattern of glucose-induced insulin release could be explained by dynamic changes in the availability of arachidonic acid and its consequent oxygenation.     

Preparation, Purification, and Stability of Tuberculin

The method used to produce “Connaught” tuberculin purified protein derivative (PPD) is described. The tuberculin PPD for the multiple-puncture method was shown to be stable for at least 24 months at 5 C; tuberculin PPD for the intracutaneous method was shown to be stable at 5 C and 24 C for a period of 18 months in the presence of Tween 80. Evans blue or brillant vital red was added to tuberculin PPD for improved testing by the multiple-puncture method. These tinted tuberculin preparations were found to be as stable as the Connaught tuberculin PPD preparations without dye at 5 C. Freeze-dried tuberculin PPD with Plasdone as an inert base was found to be remarkably stable for a period of at least 24 months at 5, 24, and 37 C.