Postprandial Glucose Dynamics and Associated Symptoms in Type 2 Diabetes Mellitus
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
KovatchevJARFall03 12/29/03 2:40 PM Page 449 Postprandial Glucose Dynamics and Associated Symptoms in Type 2 Diabetes Mellitus Boris Kovatchev, PhD* Daniel J. Cox, PhD* Kent H. Summers, PhD† Linda Gonder-Frederick, PhD* William L. Clarke, MD* *University of Virginia Health System, Charlottesville, Va † Eli Lilly & Company, Indianapolis, Ind This research was supported by Eli Lilly & Company, Indianapolis, Ind. KEY WO R D S : postprandial glucose, limit the magnitude of post-meal BG fluc- hyperglycemia, type 2 diabetes mellitus tuation may reduce its symptomatic and cognitive consequences. INTRODUCTION ABSTRACT In nondiabetic individuals postprandial glu- Objectives: This study examined the cose (PPG) fluctuations are limited in both dynamics of postprandial glucose (PPG) their peak value [rarely exceeding 7.8 and symptoms among adults with type 2 mmol/L (140 mg/dL)] and in their dura- diabetes mellitus (T2DM) in their natural tion, with a peak PPG approximately 1 environment. Using a hand-held computer hour after the start of a meal, returning to for 70 trials, 36 adults with T2DM rated preprandial levels within 2-3 hours. 1 In symptoms, performed tests, and measured individuals with type 1 diabetes (T1DM) or their blood glucose (BG). type 2 diabetes mellitus (T2DM) a number of factors, such as inadequate available Results: The mean peak PPG value was insulin, delayed insulin action, or abnor- 11.3 mmol/L achieved 2-3 hours after meal, malities in glucagon secretion, contribute while highest symptom ratings and cogni- to delayed peak PPG, and higher and pro- tive slowing were observed within the first longed PPG elevation.1 The American hour after meal, at a time corresponding to Diabetes Association Consensus Statement the steepest slope of PPG increase. Thus, on postprandial hyperglycemia concluded we hypothesized that postprandial symp- that “in general, a measurement of plasma toms maybe related to a higher rate of BG glucose 2 h after the start of a meal is prac- increase, which was confirmed by high cor- tical, generally approximates the peak value relations (r = 0.5-0.75) between symptom in patients with diabetes, and provides a ratings and BG rate of increase. reasonable assessment of postprandial Conclusions: We conclude that postprandi- hyperglycemia.” 1 al symptom elevation is related to the rate However, the dynamics of PPG are of BG increase, thus treatments designed to complex, dependent on many factors, such The Journal of Applied Research • Vol. 3, No. 4, Fall 2003 449
KovatchevJARFall03 12/29/03 2:40 PM Page 450 as the amount and the composition of the physiological as well as epidemiologic stud- meal, and still not well understood. Even ies that excessive post-load glucose excur- though a linear relationship between post- sions have acute and chronic harmful prandial and post-challenge (after a 75-g effects on the endothelium and vessel oral glucose load) glucose 2 hours after a wall.”3 Thus, an assessment of PPG dynam- meal was established in laboratory condi- ics in the natural environment would be a tions,2 the dynamics of this relationship in valuable tool for evaluation of glycemic the field is difficult to assess.3 Perhaps as a control. result of the lack of a standard PPG assess- In addition to the long-term negative ment, the usual clinical appraisal of effects of elevated PPG, clinical experience glycemic control includes only better- suggests a relationship between postprandi- defined and more stable measures, such as al hyperglycemia and acute and transient fasting plasma glucose (FPG) and/or glyco- increases in psychological symptoms and sylated hemoglobin (HbA1c). However, cognitive disruptions.2,16-18 However, there FPG reflects blood glucose (BG) values have been no prospective and objective after the effect of carbohydrate intake has investigations of the relationship of such been eliminated and HbA1c represents the symptoms/cognitive dysfunctions with post- average BG over a certain period of time, prandial BG parameters, especially with which makes both of these measures insen- parameters of postprandial BG dynamics in sitive to BG excursions throughout the the natural environment of people with dia- day—in particular to PPG fluctuations. For betes. For example, it is unclear whether example, a recent study4 of more than 800 the peak absolute value of PPG is responsi- people with T2DM found that after meals, ble for triggering symptoms, or symptoms many subjects had glucose levels >8.9 are mainly related to the speed and magni- mmol/L (160 mg/dL) and/or glucose excur- tude of BG increase post-meal. This study sions >2.2 mmol/L (40 mg/dL) despite investigates when BG peaks in T2DM HbA1c
KovatchevJARFall03 12/29/03 2:40 PM Page 451 Procedure (Lifescan, Milpitas, Calif.). Three precau- Subjects completed a series of psychomet- tions were taken to encourage and monitor ric instruments, including the Beck whether symptom entries and cognitive test- Depression Inventory. They were then ing preceded SMBG. (1) Each HHC trial instructed to use the Handspring Visor began with the message, “No blood sample Platinum, (Handspring, Inc, Mountain yet.” (2) The HHC tracked the elapsed time View, CA) hand-held computer (HHC) between the prompt “Measure your BG” immediately before self-monitoring of and the entry of this SMBG reading. Since blood glucose (SMBG). No specific SMBG at least 10 seconds are required for a sub- schedule was given to the subjects; they ject to lance a finger, collect a blood sam- were required only to complete 70 HHC tri- ple, and analyze BG level with the One als within 3-4 weeks. The HHC was Touch Ultra, any readings entered in less equipped with our custom-developed symp- than 10 seconds were considered invalid. tom/behavioral assessment software. At (3) The BG readings entered by the subjects each trial the HHC first collected data on into the HHC were compared to data in the perceived symptoms and cognitive perform- glucometer’s memory to ensure accuracy of ance. Then, subjects measured and entered SMBG results. An earlier version of this their BG level. HHC routine developed for Psion 250 HHC Symptoms. At each trial the HHC presented was used in our previous studies of symp- in a random order 16 symptoms and toms and behaviors related to hypo- prompted subjects to rate them on a scale glycemia.19-21 from 0 = none to 6 = extreme. There were 6 Data Analysis physical symptoms (need to urinate; BG values were averaged across all subjects sweet/funny taste; dry eyes, nose, mouth; in 10 time intervals post-meal and plotted tired/fatigued; thirsty; nausea), 6 mood and compared using univariate ANOVA. To symptoms (nervous/anxious; irritable/frus- obtain comparable estimates of average BG trated; restless/jittery; sad/blue; across these intervals, each time interval giddy/funny; don’t care/apathetic), and 4 was at least one half-hour in duration and cognitive symptoms (difficulty concentrat- was required to contain at least 200 SMBG ing; difficulty speaking; uncoordinated; readings, that is, at least 8% of all slowed thinking). HHC/SMBG readings. This approach Cognitive tests. The HHC presented the fol- resulted in approximately equal weights (in lowing tests: (1) 10 mental subtraction prob- terms of number of readings) of the time lems that used randomly generated 3-digit intervals. numbers, with subjects entering answers on Symptoms and cognitive test performance. a number pad; and (2) 2 levels of the Paced In order to eliminate the influence of hypo- Serial Addition Test (PSAT) presenting a glycemia, symptom ratings and test results sequence of single-digit numbers for which were considered only if BG was greater the subject has to enter the sum of each pair than 6.7 mmol/L (120 mg/dL). In order to of sequential numbers. Levels 1 and 2 of evaluate the magnitude of each individual the test present numbers at 4-second and 2- postprandial symptom its ratings were aver- second intervals, respectively. aged across subjects at 1-hour time inter- Other parameters. The HHC asked subjects vals post-meal and compared using to enter the time when “you began eating univariate ANOVA. The average symptom your last meal” and at the end of each trial, magnitude in each category, (physical, the subjects were prompted to measure and mood, and cognitive) was computed as well. enter their BG. For the latter all subjects Similarly, cognitive impairment was used One Touch Ultra glucometers assessed using the time to complete 10 The Journal of Applied Research • Vol. 3, No. 4, Fall 2003 451
KovatchevJARFall03 12/29/03 2:40 PM Page 452 Figure 1. Postprandial blood glucose (BG) levels and postprandial increase in overall symptom reporting are plotted against elapsed time after a meal (x-axis in minutes). The postprandial BG values (black squares) are presented in the primary y-axis and demonstrate a peak PPG value at approximately 2.5 – 3 hours after a meal. The number of elevated postprandial symptoms (symptoms with ratings greater than 0 over the course of PPG) is presented in the secondary y- axis (black triangles) and demonstrates a peak symptom reporting within 1 hour after a meal. mental subtractions and the number of cor- ject’s BG rate of increase (BGRI). This rect additions on the faster (Level 1) and variable is similar to the previously reported slower (Level 2) PSAT, averaged at 1-hour BG rate of change,22 but takes into account intervals post-meal. Average postprandial only increases in BG, not overall fluctua- BG 1, 2, and 3 hours after meal was corre- tions. Specifically, BGRI was computed as lated with corresponding symptom ratings the average of the ratios (BG(t2)-BG(t1))/(t2- and cognitive performance results. In order t1), where BG(t2) > BG(t1) were any two to account for multiple tests we used increasing consecutive SMBG readings of a Bonferroni corrections, accepting only subject taken at times t2 and t1 within the results significant at P 6.7 mmol/L. days). However, since fast BG increases are Rate of BG increase. From each subject’s predominantly observed after meals, we SMBG data collected concurrently with the assumed that BGRI was mostly influenced HHC, we computed an estimate of this sub- by postprandial BG elevation. 452 Vol. 3, No. 4, Fall 2003 • The Journal of Applied Research
KovatchevJARFall03 12/29/03 2:40 PM Page 453 Table 1. Average Postprandial Symptom Ratings* and Cognitive Test Performance. Pre-prandial Postprandial Ratings F (P) 1h 2h 3h Need to urinate 1.0 1.7 1.3 1.2 7.7 (
KovatchevJARFall03 12/29/03 2:40 PM Page 454 Table 2. Correlations of Average Symptom Ratings* and Cognitive Test Performance with Blood Glucose Rate of Increase. Correlation Coefficients** 1h 2h 3h Need to urinate 0.48 -0.02 0.02 Sweet/funny taste 0.39 0.01 0.08 Dry eyes, nose, mouth 0.46 -0.05 -0.02 Tired/fatigued 0.41 0.21 0.18 Thirsty 0.33 0.08 0.03 Nausea 0.42 0.02 0.12 Average magnitude of physical symptoms 0.52 0.06 0.09 Nervous/anxious 0.69 0.48 0.43 Irritable/frustrated 0.56 0.28 0.33 Restless/jittery 0.50 0.52 0.45 Sad/blue 0.68 0.53 0.44 Giddy/funny 0.50 0.07 -0.06 Not care/apathetic 0.66 0.54 0.37 Average magnitude of mood symptoms 0.70 0.49 0.44 Difficulty concentrating 0.60 0.39 0.30 Difficulty speaking 0.75 0.34 .25 Uncoordinated 0.76 0.52 .22 Slowed thinking 0.58 0.34 .31 Average magnitude of cognitive symptoms 0.74 0.44 0.30 Time to complete 10 mental subtractions (sec.) 0.27 0.06 0.02 PSAT – Level 1 Number correct answers -0.21 -0.21 0.18 PSAT – Level 2 Number correct answers -0.26 -0.02 0.15 * Using a scale from 0 = none to 6 = extreme. ** With this sample size correlations above 0.37 yield P-levels below P= 0.05, while correlations above 0.47 are significant at P = 0.01. The latter are indicated in bold. Postprandial Symptoms and Cognitive (Table 1). Slowing Specifically, 5 of the 6 physical symp- Most physical, mood, and cognitive symp- toms were rated higher post-meal: subjects toms displayed similar patterns of highest reported a greater need to urinate (F = 7.7, average rating within the first hour after P
KovatchevJARFall03 12/29/03 2:40 PM Page 455 thirst (F = 11.2, P
KovatchevJARFall03 12/29/03 2:40 PM Page 456 tions with BGRI were highest within 1 hour design of the study was that the BG rate of post-meal, and gradually decreased at 2 and increase could not be computed strictly for 3 hours post-meal. Since PPG increase pre-to-postprandial periods, that is, the slowed down at 2 and 3 hours post-meal, slope of PPG increase could not be estimat- and the PPG curve gradually flattened, this ed directly. This was due to the lack of suf- was precisely the effect to be expected if ficient number of BG readings within a symptoms were related to PPG increase and pre- and corresponding postprandial peri- not to extreme PPG values. With this sam- od; at least two, pre-and postprandial, read- ple size (N = 36) correlations above 0.47 ings on several days per subject are needed yielded P-levels below 0.01 and were con- in order to estimate PPG slope. sidered significant. Several correlation Instead, we confirmed a less specific coefficients were above 0.6. For example, hypothesis: higher postprandial symptoms the average magnitude of “difficulty speak- are related to higher overall BG rate of ing” and “uncoordinated” 1 hour post-meal increase. The BGRI was based on a previ- produced a correlation coefficient of 0.75- ously reported measure, BG rate of 0.76 with BGRI (see Table 2). change,22 but it took into account only con- secutively increasing SMBG readings, not DISCUSSION any two consecutive readings. As computed This study used new monitoring tech- here, BGRI was not specific to postprandial nology (HHC and custom software) to time periods; however, we assumed that it assess in-the-field dynamics of postprandial was most influenced by the largest (per BG and associated symptoms and cognitive hour) BG increases that subjects experi- impairment in adults with type 2 diabetes enced post-meal. Indeed, it was illogical to mellitus. Our data allowed for a reconstruc- expect that large and fast BG elevations tion of PPG dynamics in our subjects’ natu- could have occurred with no relationship to ral environment and confirmed laboratory a preceding significant carbohydrate intake. observations and common knowledge that Thus, we could speculate that the process of PPG is most elevated 2-3 hours after meal. 1 PPG elevation contributed to symptoms In addition, we demonstrated that after more than the absolute values of postpran- meals self-reported physical symptoms and dial hyperglycemia. While this speculation moods became significantly elevated, and may be confirmed (or rejected) by future that objectively determined cognitive slow- studies, here we were able to clearly ing of approximately 30% was apparent. observe two related properties: (1) symp- Surprisingly, however, symptom eleva- tom ratings were higher during the first tion and cognitive slowing did not follow hour post-meal and this was the period of the course of absolute PPG values. Instead, steepest slope of PPG increase, and (2) for most symptoms highest symptom rat- postprandial symptoms were highly corre- ings were observed within 1 hour after a lated with subject’s overall rate of BG meal (Table 1), not when PPG was at its increase. peak 2-3 hours post-meal. Overall, plotting In fact, the correlations between mood a summary of all symptoms and PPG and cognitive symptoms and BGRI were against the time elapsed after a meal very high, in some cases higher than 0.7 revealed that symptoms generally occurred (Table 2). Given that symptom ratings were during the time of steepest slope of PPG derived from behavioral self-assessment, increase (Figure 1). Thus, we hypothesized while BGRI was derived from concurrent that postprandial symptoms occurred during but quite different sets of SMBG data times when BG increase was fastest. With downloaded from the subjects’ glucometer our data, however, we could not definitely memories, correlations of that magnitude confirm this notion. One limitation in the imply very strong [linear] relationships. 456 Vol. 3, No. 4, Fall 2003 • The Journal of Applied Research
KovatchevJARFall03 12/29/03 2:40 PM Page 457 Thus, the magnitude and speed of BG 2. Wolever TMS, Palmason C, Chiasson J, et al. Variation of postprandial plasma glucose, palata- increase post-meal may be the single most bility, and symptoms associated with a standard- significant determinant of postprandial ized mixed test meal versus 75 g oral glucose. symptoms. Diabetes Care. 1998;21:336-340. In contrast to symptoms, reduced per- 3. Hanefeld M. Postprandial hyperglycemia: nox- ious effects on the vessel wall. Int J Clin Pract. formance on cog n i t ive tests, although well 2002;129(suppl):45-50. expressed in the first hour after meal (Table 4. Bonora E, Calcaterra F, Lombardi S, Bonfante N, 1, cog n i t ive tests), could not be explained by et al. Plasma glucose levels throughout the day absolute PPG peak, or by BGRI. Thus, alter- and hba1c interrelationships in type 2 diabetes: native mechanisms of postprandial cognitive implications for treatment and monitoring of metabolic control. Diabetes Care. 2001;24:2023- slowing must be considered. For example, 2029. there may be a BG threshold, above which 5. Avignon A, Radauceanu A, Monnier L. these effects occur in a stepwise fashion, or Nonfasting plasma glucose is a better marker of some disruption of metabolic homeostasis diabetic control than fasting plasma glucose in may be responsible for the observed cog n i- type 2 diabetes. Diabetes Care. 1997;20:1822- 1826. tive slowing. Further studies will be needed 6. Caputo S, Pitocco D, Ruotolo V, Ghirlanda G. to address these issues. What is the real contribution of fasting plasma The results of this study imply that in glucose and postprandial glucose in predicting people with T2DM the most effective time HbA1c and overall blood glucose control? Diabetes Care. 2001;24:2011-2011. to sample symptoms and cognitive func- 7. The Diabetes Control and Complications Trial tioning postprandially is during the first 1-2 Research Group. The effect of intensive treat- hours after a meal. This is also the time ment of diabetes on the development and pro- when fastest BG elevation was observed in gression of long-term complications in insulin-dependent diabetes mellitus. N Engl J this study and documented by others.0 Since Med. 1993;329:977–986. the BGRI was a better correlate of post- 8. UK Prospective Diabetes Study (UKPDS) prandial symptoms than absolute PPG Group. Intensive blood-glucose control with peaks, treatment regiments designed to sulphonylureas or insulin compared with conven- reduce the magnitude and the speed of PPG tional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). fluctuations, such as rapid-acting insulin Lancet. 1998;352:837–853. analogs or complex carbohydrate diets, may 9. Hanefeld M, Temelkova-Kurktschiev T. The post- reduce the negative symptomatic and cogni- prandial state and the risk of atherosclerosis. tive consequences of postprandial glucose Diabetic Med. 1997;14(suppl 3): S6-11. excursions. 10. Soonthornpun S, Rattarasarn C, Leelawattana R, Setasuban W. Postprandial plasma glucose: a good index of glycemic control in type 2 diabetic ACKNOWLEDGMENTS patients having near-normal fasting glucose lev- The authors would like to thank els. Diabetes Res Clin Pract. 1999:46:23–27. Pamela Erickson, Haya Ascher-Savanum, 11. Gavin JR III. The importance of postprandial PhD, and James Malone, MD, for their hyperglycaemia. Int J Clin Pract. 1999;107 (suppl):14-17. helpful reviews and suggestions on previous drafts of this manuscript. 12. Hanefeld M. Post-prandial hyperglycaemia and vascular disease. Int J Clin Pract. 2000; 112(suppl):13-18. REFERENCES 13. Rajmohan L, Mohan V, Ramanujam TR. 1. American Diabetes Association. Postprandial Postprandial hyperglycaemia—the real challenge Blood Glucose: Consensus Statement. Diabetes in diabetes. J Assoc Physicians India. 2001; Care. 2001; 24:775-778.2. Wolever TMS, 49:357-360. Palmason C, Chiasson J, et al. Variation of post- prandial plasma glucose, palatability, and symp- 14. Haffner S. The importance of postprandial hyper- toms associated with a standardized mixed test glycaemia in development of cardiovascular dis- meal versus 75 g oral glucose. Diabetes Care. ease in people with diabetes. Int J Clin Pract. 1998;21:336-340. 2001;123(suppl):24-26. The Journal of Applied Research • Vol. 3, No. 4, Fall 2003 457
KovatchevJARFall03 12/29/03 2:40 PM Page 458 15. Hanefeld M, Fisher S, Julius U. Risk Factor for 19. Kovatchev BP, Cox DJ, Gonder-Frederick LA, et myocardial infarction and death in newly detect- al. Stochastic model of self-regulation decision ed NIDDM: the Diabetes Intervention Study, 11- making exemplified by decisions concerning year follow-up. Diabetologia. hypoglycemia. Health Psychol. 1998;17:277-284. 1996;39:1577-1583. 20. Cox DJ, Gonder-Frederick LA, Kovatchev BP, et 16. Van der Does FE, De Neeling JN, Snoek FJ, et al: Biopsychobehavioral model of severe hypo- al. Symptoms and well-being in relation to glycemia. II. Understanding the risk of severe glycemic control in type II diabetes. Diabetes hypoglycemia. Diabetes Care. 1999;22: 2018- Care. 1996;19:204-210. 2025. 17. De Sonnaville JJ, Snoek FJ, Colly LP, et al. Well- 21. Kovatchev BP, Cox DJ, Robeva RS, et al. being and symptoms in relation to insulin thera- Quantifying bio-behavioral determinants of risk py in type 2 diabetes. Diabetes Care. for severe hypoglycemia in Type 1 diabetes. J 1998;21:919-924. Appl Res. 2001;1:16-23. 18. Cox DJ, Gonder-Frederick LA, McCall A, et al. 22. Kovatchev BP, Cox DJ, Gonder-Frederick LA, The effects of glucose fluctuation on cognitive WL Clarke. Methods for quantifying self-moni- function and QOL: the functional costs of hypo- toring blood glucose profiles exemplified by an glycaemia and hyperglycaemia among adults examination of blood glucose patterns in patients with type 1 or type 2 diabetes. Int J Clin Pract. with Type 1 and Type 2 Diabetes. Diabetes 2002;129(suppl):20-26. Technol Ther. 2002; 4:295-303. 458 Vol. 3, No. 4, Fall 2003 • The Journal of Applied Research
You can also read