Original ArticleArticle original

AGING EXAGGERATES THE BLOOD GLUCOSE RESPONSETO TOTAL PARENTERAL NUTRITION

James M. Watters, MD, FRCSC;* Susan M. Kirkpatrick, RN; Donna Hopbach, RN; Sonya B. Norris, MSc

From the Department of Surgery, University of Ottawa, and Loeb Research Institute, Ottawa Civic Hospital, Ottawa, Ont.

Supported by Medical Research Council of Canada Grant MT-10030

*Supported by a Career Scientist Award, Ontario Ministry of Health

Accepted for publication July 12, 1996

Correspondence to: Dr. James M. Watters, Ottawa Civic Hospital, 1053 Carling Ave., Ottawa ON K1Y 4E9; fax 613 761-5371

© 1996 Canadian Medical Association (text and abstract/résumé)

OBJECTIVE: To determine the effect of age on the blood glucose and insulin responses to a clinical modelof glucose loading (i.e., total parenteral nutrition [TPN] with hypertonic glucose), in patients with a vari-ety of conditions.DESIGN: A prospective cohort study.SETTING: An adult university hospital.PATIENTS: Seventy-one consecutive, clinically stable patients receiving central TPN, excluding those withmetabolic disease or receiving relevant medications.INTERVENTION: None.MAIN OUTCOME MEASURES: Serum levels of glucose, insulin, C-peptide and cortisol determined in periph-eral venous blood obtained immediately before initiating TPN and again 48 to 96 hours later; acute physi-ology score (APS) and habitual level of physical activity (HAL). RESULTS: Serum levels of glucose, insulin and C-peptide increased following initiation of TPN (allp < 0.001). The serum glucose level during TPN administration increased as a function of both patient ageand severity of illness (APS) (r2 = 0.37, all p < 0.01), whereas the serum insulin level was inversely relatedto age and increased as a function of serum glucose, glucose rate of infusion and HAL (r2 = 0.57, allp < 0.05). The serum C-peptide:insulin molar ratio did not vary with age.CONCLUSIONS: Aging and severity of illness interact to exaggerate the increases in blood glucose that ac-company TPN with hypertonic glucose. Serum insulin responses to TPN decline with aging, likely reflect-ing reduced insulin secretion. Diminished insulin responses may contribute to hyperglycemia and representa diminished anabolic signal in such patients. The acutely ill elderly patient is predisposed to hyperglycemiaand should be monitored carefully even when pre-TPN blood glucose values are normal.

OBJECTIF : Déterminer l’effet de l’âge sur les réactions de la glycémie et de l’insuline à un modèle cliniquede charge de glucose (c.-à-d. nutrition parentérale totale [NPT] avec glucose hypertonique) chez les pa-tients atteints de toutes sortes d’affections.CONCEPTION : Étude prospective de cohortes.CONTEXTE : Hôpital universitaire pour adultes.PATIENTS : Soixante et onze patients consécutifs, cliniquement stables, qui ont reçu une NPT centrale. Ona exclu ceux qui avaient des problèmes du métabolisme ou qui prenaient des médications pertinentes.INTERVENTION : Aucune.PRINCIPALES MESURES DES RÉSULTATS : Taux sériques de glucose, d’insuline, de peptide C et de cortisonedéterminés dans le sang veineux périphérique prélevé immédiatement avant le début d’une NPT et, de nou-veau, de 48 à 96 heures plus tard; score physiologique de gravité (SPG) et niveau habituel d’activité physique.

14459 December/96 CJS /Page 481

CJS, Vol. 39, No. 6, December 1996 481

Normal aging is accompaniedby predictable changes inglucose homeostasis, which

include minor increases in fastingblood glucose levels, impaired toler-ance of glucose loads and diminishedtissue sensitivity to insulin.1,2 Changesin carbohydrate metabolism are alsocentral among the metabolic responsesto acute surgical illness; they includehyperglycemia, glucose intolerance andresistance of normally insulin-sensitivetissues to its effects.3 We have observedthat increases in fasting blood glucoselevels after trauma are exaggerated inolder patients and that such patients aremarkedly intolerant of short-term (2-hour) glucose loads.4,5 To determinewhether comparable age effects occurin a clinical model of glucose loading(i.e., total parenteral nutrition [TPN]with hypertonic glucose administeredover a period of days), we evaluated theserum glucose, insulin and C-peptideresponses to TPN in patients hospital-ized with a variety of illnesses.

PATIENTS AND METHODS

Consecutive, clinically stable pa-tients for whom central venous TPNwith hypertonic (20%) glucose wasplanned as the sole nutrient sourcewere eligible for study. Those with ahistory of diabetes mellitus, pregnant,or receiving corticosteroids or ß-adrenergic blocking agents were ex-cluded. Severity of illness was evalu-ated in terms of acute physiology

score (APS) and ward versus intensivecare unit (ICU) location.6 The refer-ence period for the APS was the24 hours immediately before initialblood sampling. Habitual level ofphysical activity (HAL) before hospi-talization was assessed according to astandardized descriptive scale.7

Peripheral venous blood was ob-tained immediately before the start ofTPN and again 48 to 96 hours afterfor determination of the serum glucoselevel, by glucose oxidase methodology(Glucose Analyzer 2; Beckman Instru-ments, Palo Alto, Calif.) and serum in-sulin, C-peptide and cortisol levels byradioimmunoassay (Euro/DPC, Wit-ney, UK). TPN was provided as 4.25%or 7% amino acids (Aminosyn II 7%;Abbott Laboratories, Saint-Laurent,Que.) in 20% dextrose and lipid emul-sion 10% or 20% (Liposyn II 20%; Ab-bott Laboratories). The rates of infu-sion of glucose, lipid and protein at thetime of the second blood samplingwere recorded, together with specificdiagnoses, information on the admin-istration of insulin and other clinicaldata.Data are given as means (and stan-

dard deviations) and were analysed byunpaired and paired t-tests, the χ2 testwith Yates’ correction and linear re-gression (SystatFPU 5.2.1; Systat Inc.,Evanston, Ill.). The protocol was ap-proved by the Research Ethics Com-mittee of the Ottawa Civic Hospital,and written consent was obtainedfrom each patient or next of kin.

FINDINGS

Seven of 78 patients studied re-ceived exogenous insulin after thestart of TPN. They were similar in ageto those who did not receive insulinbut had higher blood glucose concen-trations before initiation of TPN (p <0.001). All patients who received ex-ogenous insulin were in an ICU (p <0.05). These seven patients were ex-cluded from further analysis. The re-maining 71 patients (42 men, 29women) ranged in age from 18 to83 years (Fig. 1). Diagnoses were het-erogeneous (Table I), and the APSranged from 0 to 30. Thirty-two pa-tients were in an ICU at the time ofstudy. They had a higher APS (p <0.001) and serum cortisol level (p <0.05) than the 39 ward patients, buttheir serum glucose and insulin levelswere not different. The mean age ofpatients in the ICU was also greater(62 [16] v. 51 [17] years, p < 0.01),but there was considerable overlap in

WATTERS ET AL

14459 December/96 CJS /Page

482 JCC, Vol. 39, No 6, décembre 1996

RÉSULTATS : Les taux sériques de glucose, d’insuline et de peptide C ont augmenté après le début de laNPT (tous p < 0,001). Le taux de glucose sérique au cours de l’administration de la NPT a augmenté enfonction à la fois de l’âge du patient et de la gravité de la maladie (SPG) (r2 = 0,37, tous p < 0,01), tandisque le taux d’insuline sérique présentait une relation inverse avec l’âge et a augmenté en fonction du tauxde glucose sérique, du taux d’infusion du glucose et du niveau habituel d’activité physique (r2 = 0,57, tousp < 0,05). Le ratio molaire peptide C:insuline sérique n’a pas varié en fonction de l’âge.CONCLUSIONS : Le vieillissement et la gravité de la maladie interagissent pour aggraver les hausses de la gly-cémie qui accompagnent une NPT avec glucose hypertonique. Les réactions de l’insuline sérique à la NPTdiminuent avec le vieillissement, ce qui reflète probablement une baisse de la sécrétion d’insuline. Des réac-tions réduites de l’insuline peuvent contribuer à une hyperglycémie et représentent une baisse du signal ana -bolique chez de tels patients. Le patient âgé gravement malade est prédisposé à l’hyperglycémie et il faut lesuivre attentivement, même lorsque les valeurs de la glycémie avant la NPT sont normales.

FIG. 1. Age distribution of patients (5-year inter-vals).

Patient age, yr

ages, and age ranges were similar (25to 83 v. 18 to 82 years respectively).The relationship between age andseverity of illness (APS), although sig-nificant (p < 0.05), was very limited (r2= 0.08). Patient age and HAL wereweakly related (r2 = 0.17, p < 0.05),the HAL tending to decline with age.Serum glucose levels before the ini-

tiation of TPN did not vary with pa-tient age, APS, HAL, weight, bodymass index or the rate of glucose ad-ministered in pre-TPN intravenoussolutions. Serum insulin levels weresomewhat related to HAL (r2 = 0.20,p < 0.05), tending to increase with anincreased level of physical activity.Serum C-peptide levels and C-pep-tide:insulin molar ratios increased withthe severity of illness (APS) (r2 = 0.21and 0.25, respectively, both p <

0.001), but neither was a function ofage. Serum cortisol levels increasedwith increasing age and with dimin-ished HAL (r2 = 0.38, all p < 0.01). The interval from the initiation of

TPN to the second blood samplingdid not vary with patient age. The rateof TPN administration at that timewas equivalent to 1540 (383) nonpro-tein calories per 24 hours (25 [10]kcal/kg every 24 hours, range 700 to2440 kcal/kg every 24 hours), com-prising glucose 920 (250) kcal/24 hand lipid 620 (245) kcal/24 h, to-gether with protein 57 (16) g/24 h(1.0 [0.4] g/kg every 24 hours). Pro-tein, total nonprotein calories, andglucose and lipid calories adminis-tered, expressed either in absoluteterms or as a function of body weight,did not vary with patient age. Patients

in the ICU were receiving an averageof 20% less glucose at the time of thesecond blood sampling than ward pa-tients (p < 0.01) and a greater propor-tion of nonprotein calories as lipid(p < 0.05).Serum glucose, insulin and C-

peptide levels increased after TPN wasbegun, whereas serum cortisol levelsdid not (Table II). None was influ-enced by the duration of TPN admin-istration before blood sampling.Serum glucose levels during TPN ad-ministration increased as a function ofage (p < 0.05) and APS (p < 0.01),but were most strongly related to theage–APS interaction and pre-TPNglucose levels (r2 = 0.37, all p < 0.01)(Fig. 2). Serum glucose levels werenot influenced by the rate of glucoseinfusion. Although not related to ageon univariate analysis, serum insulinlevels during TPN administration de-clined with increasing patient age andincreased as a function of the serumglucose level, rate of infusion of glu-cose and HAL (r2 = 0.57, all p < 0.05)on multivariate analysis. Serum insulinlevels were not related to the rate ofamino acid infusion or serum cortisollevel. Serum C-peptide levels in-creased with APS (r2 = 0.27, p <0.001), but neither C-peptide levels

AGING AND BLOOD GLUCOSE RESPONSE TO TPN

14459 December/96 CJS /Page 483

CJS, Vol. 39, No. 6, December 1996 483

Table II

(187)

(2490)*

(217)*

(2.8)*

Measurements of Serum Levels Before and During TPN(Means [and Standard Deviations])

Measurement

Glucose, mmol/L

Insulin, pmol/L

C-peptide, nmol/L

Cortisol, nmol/L

*p < 0.001 v. before TPN

558

1310

80

5.9

Before TPN

(276)

(1270)

(79)

(1.4)

518

2820

261

7.5

During TPN

Table I

Clinical Settings in Which 71 Patients Received Total Par-enteral Nutrition (TPN)

Clinical setting

Ileus after abdominal sepsis or surgery

Multiple organ dysfunction

Fistula (intestinal, pancreatic, lymph)

Acute pancreatitis

Multiple trauma including abdomen 5

8

13

15

14

No. of patients

Bone marrow transplant 4

Other (enteritis, colitis, short-bowelsyndrome) 12

Total 71

FIG. 2. Serum glucose concentrations after initi-ation of total parenteral nutrition, as function ofage in patients with heterogeneous diagnosesand varying severities of illness.

nor C-peptide:insulin molar ratios var-ied with patient age.

DISCUSSION

Carbohydrate metabolism aftertrauma and during sepsis has beenwell characterized in previouslyhealthy young and middle-aged pa-tients: hyperglycemia is usual despitenormal or elevated serum insulin lev-els, endogenous glucose production isaccelerated, tissues that are normallysensitive to the effects of insulin be-come relatively insensitive and toler-ance of glucose loads is diminished.8–11

The magnitude of these changes tendsto reflect the severity of the stress, atleast after injury,4,12 and we observedthat serum glucose levels during TPNincreased with the severity of illness inpatients with a wide variety of diag-noses.Generally predictable changes in

glucose metabolism accompany nor-mal aging as well. Age-related changesinclude minor increases in fastingblood glucose levels and impaired tol-erance of glucose loads, and they in-teract with those that follow injury.1,2

We have observed in previous workthat aging was associated with increas-ing fasting serum glucose levels dur-ing both the initial “ebb” and thelater, hypermetabolic “flow” phases ofthe injury response.4 We have alsodemonstrated a substantial reductionin the tolerance of an acute (2-hour)glucose load in older trauma patients,associated with markedly reduced in-sulin responses.5 The present observa-tions extend those of earlier studies inthat responses to glucose loading in aclinical form (i.e., TPN) were evalu-ated over longer periods (at least48 hours) in patients hospitalized witha variety of acute conditions. Serumglucose was found to increase as afunction of both patient age andseverity of illness. This relationship

was noted despite the considerablerange of diagnoses, the limited screen-ing of patients for diabetes and otherchronic conditions, the exclusion ofpatients who received exogenous in-sulin (i.e., those who were most hy-perglycemic) and the heterogeneouseffects of aging itself.13 Moreover, nodirect account was taken of the manyfactors that potentially influence glu-cose tolerance such as bed rest, renalinsufficiency and administration ofmorphine, although they are presum-ably reflected in part in the APS andlocation (ward v. ICU).14,15 It is alsolikely that patients in whom glucoseintolerance and hyperglycemia wereanticipated were administered smallerglucose loads; for example, patients inthe ICU were receiving somewhat lessglucose at the time of the secondblood sampling than were ward pa-tients. Despite these sources of vari-ability, blood glucose responses toTPN were significantly influenced bypatient age and severity of illness, andparticularly by the interaction of thesetwo factors.Glucose is the most important

stimulus to insulin secretion, andprompt and sustained increases in pe-ripheral serum insulin levels duringparenteral nutrition with hypertonicglucose have long been recog-nized.16,17 Insulin responses to hyper-glycemia are preserved in healthy, un-stressed older people, but they arealtered by stresses such as injury.2,5

The insulin responses of older traumapatients to acute glucose loading havebeen shown to be markedly decreasedcompared with those of young pa-tients.5 The inverse relationship be-tween age and serum insulin duringTPN administration in this study isconsistent with such observations, andage-related increases in blood glucoselevels may have been, in part, a resultof lower serum insulin concentrations.The ages of patients who received ex-

ogenous insulin were similar to thosewho did not; that older patients werenot overrepresented among the for-mer may reflect the small number ofsuch patients, that all seven were in anICU, where surveillance may be morefrequent and insulin administered at alower threshold, or that severity of ill-ness is a more powerful determinantof hyperglycemia than age. Whetheraging influences the insulin resistancethat follows injury and other acutestresses also remains to be determined.The elderly patient is also predisposedto hyperglycemia by an increasedthreshold for the renal clearance ofglucose,18 but urine glucose levelswere not measured in this study.The age-related decline in the

serum insulin level that we observedmay have resulted from decreasedpancreatic secretion of insulin or in-creased clearance from the circulation.C-peptide is a byproduct of insulinbiosynthesis, co-secreted from thepancreatic beta cell on an equimolarbasis with insulin, and a marker of in-sulin secretion.19 The absence of anage effect on C-peptide:insulin molarratio during TPN administration isconsistent with diminished insulin se-cretion (rather than accelerated clear-ance) as the explanation for lowerserum insulin levels in older patients,although it may simply reflect a lackof statistical power. A number of fac-tors other than serum glucose influ-ence insulin secretion. Insulin secre-tion is generally stimulated by aminoacids and in settings of hypercorti-solemia, but we did not observe statis-tical relationships between serum in-sulin and the rate of amino acidinfusion or serum cortisol.19 We didobserve a modest, direct relationshipbetween habitual physical activity andserum insulin levels, both before andduring TPN infusion. This contrastswith observations in healthy subjects,in whom higher levels of habitual

WATTERS ET AL

14459 December/96 CJS /Page

484 JCC, Vol. 39, No 6, décembre 1996

physical activity are associated with in-creased tissue sensitivity to insulin anddecreased insulin responses to glu-cose.20–22 However, bed rest for peri-ods of several days or more is accom-panied by significant increases ininsulin levels and diminished glucosetolerance.14,23 The relationships amonginsulin secretion and clearance, tissuesensitivity to insulin and physical ac-tivity are likely to be considerably per-turbed by the combined effects of ageand acute illness. For example, thechange to very limited physical activ-ity that accompanies illness and hospi-talization may be associated with moremarked increases in serum insulin inpreviously active individuals.In summary, blood glucose levels

during TPN increased as a function ofage in a heterogeneous group of hos-pitalized patients, whereas serum in-sulin responses decreased. Careful sur-veillance for hyperglycemia afterinitiating TPN is essential in the el-derly even when previous blood glu-cose values have been within the nor-mal range and especially in the mostaged and most acutely ill patients. Hy-perglycemia should be controlled withexogenous insulin, and considerationshould be given to the use of lipid toprovide an increased proportion ofnonprotein calories in such patients.The diminished anabolic signal repre-sented by impaired insulin responsesin the elderly may also be disadvanta-geous, since the muscle mass andstrength of the elderly patient are pre-dictably reduced even before acute ill-ness or injury.

References

1. Davidson MB. The effect of aging oncarbohydrate metabolism. Metabo-lism 1979;28:688-705.

2. DeFronzo RA. Glucose intolerance andaging: evidence for tissue insensitivity

to insulin. Diabetes 1979;28:1095-101.

3. Bessey PQ, Watters JM, Brooks DR,Black PR, Wilmore DW. Glucose me-tabolism following trauma or sepsis.In: Clowes GH Jr, editor. Trauma,sepsis and shock: the physiologic basis oftherapy. New York: Marcel Dekker,1988.

4. Desai D, March RJ, Watters JM. Hyperglycemia following trauma in-creases with age. J Trauma 1989;29:719-23.

5. Watters JM, Moulton SB, ClanceySM, Blakslee JM, Monaghan R. Ag-ing exaggerates glucose intolerancefollowing injury. J Trauma 1994;37:786-91.

6. Knaus WA, Draper EA, Wagner DP,Zimmerman JE. APACHE II: a sever-ity of disease classification system.Crit Care Med 1985;13(10):818-29.

7. Grimby G. Physical activity and muscletraining in the elderly [review]. ActaMed Scand Suppl 1986;711:233-7.

8. Wilmore DW, Mason AD Jr , PruittBA Jr. Insulin response to glucose inhypermetabolic burn patients. AnnSurg 1976;183:314-20.

9. Black PR, Brooks DC, Bessey PQ,Wolfe RR, Wilmore DW. Mecha-nisms of insulin resistance followinginjury. Ann Surg 1982;196:420-35.

10.Allison SP, Hinton P, Chamberlain MJ.Intravenous glucose tolerance, insulinand free fatty acid levels in burned patients. Lancet 1968;2:1113-6.

11. Shangraw RE, Jahoor F, Miyoshi H,Neff WA, Stuart CA, Herndon DN,et al. Differentiation between septicand postburn insulin resistance. Me-tabolism 1989;38(10):983-9.

12. Howard JM. Studies of absorptionand metabolism of glucose followinginjury. Ann Surg 1955;141:321-6.

13. Rowe JW, Kahn RL. Human aging:usual and successful. Science 1987;237:143-9.

14. Mikines KJ, Dela F, Tronier B, Galbo

H. Effect of 7 days of bed rest on dose-response relation between plasma glu-cose and insulin secretion. Am J Phys-iol 1989;257(1 Pt 1):E43-8.

15. Radosevich PM, Williams PE, LacyDB, McRae JR, Steiner KE, Cher-rington AD, et al. Effects of mor-phine on glucose homeostasis in theconscious dog. J Clin Invest 1984;74(4):1473-80.

16. Genuth SM. Insulin response to in-travenous alimentation [letter]. NEngl J Med 1973;289:107.

17. Sanderson I, Deitel M. Insulin responsein patients receiving concentrated infu-sions of glucose and casein hydrolysatefor complete parenteral nutrition. AnnSurg 1974;179:387-94.

18. Butterfield WJ, Keen H, WhichelowMJ. Renal glucose threshold varia-tions with age. Br Med J 1967;4:505-7.

19. Polonsky KS, Rubenstein AH. Thekinetics and metabolism of insulin,proinsulin, and C-peptide. In: De -Groot LJ, Besser GM, Cahill GF Jr,editors. Endocrinology. 2nd edition.Philadelphia: W.B. Saunders Com-pany, 1989: 1304-17.

20. Kahn SE, Larson VG, Beard JC, CainKC, Fellingham GW, Schwartz RS, etal. Effect of exercise on insulin action,glucose tolerance, and insulin secre-tion in aging. Am J Physiol 1990;258(6 Pt 1):E937-43.

21. Soman VR, Koivisto VA, Deibert D,Felig P, DeFronzo RA. Increased in-sulin sensitivity and insulin bindingto monocytes after physical training.N Engl J Med 1979;301:1200-4.

22. Wasserman DH, Geer RJ, Rice DE,Bracy D, Flakoll PJ, Brown LL, et al.Interaction of exercise and insulin ac-tion in humans. Am J Physiol 1991;260(1 Pt 1):E37-45.

23. Rogers MA, King DS, Hagberg JM,Ehsani AA, Holloszy JO. Effect of 10days of physical inactivity on glucosetolerance in master athletes. J ApplPhysiol 1990;68:1833-7.

AGING AND BLOOD GLUCOSE RESPONSE TO TPN

14459 December/96 CJS /Page 485

CJS, Vol. 39, No. 6, December 1996 485