N A L A R T I C L

Comparison of Different Insulin Regimensin Elderly Patients With NIDDMBRUCE H.R. WOLFFENBUTTEL, MD, PHDJl-AN-PlFRRt J.E. SHLS, MD, PHDGABRIELLEJ.W.M. RONDAS-COLBERS

PAUL P.C.A. MENHEERE, PHDARtE C . NlEUWENHUIJZEN K R U S E M A N , MD, PHD

OBJECTIVE — To compare the metabolic effects of three different frequently used regi-mens of insuliti administration on blood glucose control and serum lipids, and the costs asso-ciated with this treatment, in .subjects with NIDDM, who were poorly controlled with oralantihyperglycemic agents.

RESEARCH DESIGN AND METHODS— We studied 95 elderly patients with NIDDM(age 68 ± 9 years, BMI 26.0 ± 4.6 kg/m-, and tnedian time since diagnosis of diabetes 9 yearsIrange 1-371; 37 men, 58 women), who were poorly controlled, despite diet and maximaldoses of oral antihyperglycemic agents. Three insulin administration regimens were comparedduring a 6-month period: patients were randomized for treatment with a two-injection scheme(regimen A) or a combination of glibenclamide with one injection of NPH insulin, adminis-tered either at bedtime (regimen B) or before breakfast (regimen C), and insulin treatment wasmainly instituted in an outpatient setting.

RESULTS — After 6 months of insulin treatment, fasting blood glucose of the total patientpopulation had decreased from an average of 14.1 ± 2.2 to 8.3 ± 2.0 mmol/1 (P < 0.001), andHbA,, fell from 11.0 ± 1.3 to 8.3 ± 1.2% (P < 0.001); 34 patients reached HbAi, levels below8.0%, 25 of them even below 7.5%. With two insulin injections daily, HhAi decreased from11.2 ± 1.3 to 8.2 ± 1.2%, while during combined treatment, HbA^. fell from 10.5 ± 1.2 to 8.1± 1.1% (regimen B) and from 11.1 ± 1.3to8.5± 1.1% (regimen C). Comparable improvementof the other measures of glycémie control, lipids and lipoproteins, was observed in the differ-ent treatment regimens. Body weight increase was rnoderate (mean ± 4.0 kg) and similar in allpatient groups. One-third of patients starting with one insulin injection daily needed a secondinjection to control glycemia. One episode of severe hypoglycemia was observed. Combinedinsulin-sulfonylurea treatment was almost 20% more expensive than twice-daily administra-tion of insulin alone.

CONCLUSIONS— Insulin treatment can safely be instituted in elderly patients withNIDDM. However, it is difficult to obtain optimal glycémie control. Insulin has moderatebeneficial effects on seritm lipoproteins. Although on the basis of glycémie control and weightgain, no preference for any treatment regitnen cati he discerned, twice-daily insulin adminis-tration is the most simple and cost-effective regimen.

Patients with NIDDM are usuallytreated with oral antihyperglycemicagents when diet therapy alone fails

10 tiormalize glucose metalDolism (1,2).For patients not achieving adequate con-trol with such therapy, insulin treatment isthe available option (3-5); the aim of this

treatment is to diminish hyperglycémiecomplaints, improve blood glucose con-trol, and treat plasma Upid disturbances(6,7) to improve quality of life and preventthe development or worsening of diabeticcomplications. Weight gain during insulintherapy may pose a problem, especially in

liom the LX'partmenl ol Inlcrnal Medicine. Division ol Endocrinology, and Department oí Clinical Chem-istry' (PPC.A.M.). University Hospital Maaslricht, Maastiicht. The Netherlands.

Address correspondence and reprint requests to Bruce H.R. Wolffenbutlel, MD, PhD, Dept. of InternalMedicine. Division of Endocrinology and Metabolism, tjniversity Hospital Maastricht. PO. Box 5800, NL-6202 AZ Maastricht. The Netherlands. E-mail: bwo@sint.azm.nl.

Received for publication 6 February 1996 and accepted in revised form 11 July 1996,apo, apolipoprotein; AUC, area under the curve; AUCFBG, ratio of C-peptide AUC and fasting blood

glucose; Lp(a). lipoprotein(a); RIA. radioimmunoassay; UGDP, University Group Diabetes Project.

Líbese patients. Further, the resultingperipheral hyperinsulinemia might be ofimportance, since some authors (8,9) con-sider increased plasma insulin levels as anindependent risk factor for the develop-ment of atherosclerosis. In addition, theincidence of hypoglycemia during insulintherapy is reported to be higher than dur-ing oral treatment (10).

Various insulin regimens have beenproposed. During recent years, the interestin the combination of insulin with sul-fonylureas has revived (11). This combi-nation may be beneficial by reducing thedose of insulin needed and, thus, possiblydiminishing the degree of peripheralhyperinsulinemta. The injection of inter-mediate-acting insulin at bedtime,together with daytime sulfonylurea use,was especially advocated (12-14) to sup-press the excessive nocturnal hepatic glu-cose production. This regimen trtay alsoprovoke less weight gain (14), but nega-tive effects of combined insulin-sulfony-lurea therapy on HDL cholesterol havebeen described (15).

The purpose of thts study was toassess the metabolic effects of insulin ther-apy on blood glucose control and serumlipids, and the costs associated with thistreatment, in subjects with NIDDM, whowere poorly controlled with maximaldoses of oral agents. Three different fre-quently advocated regimens of insulinadministration, alone or combined withsulfonylureas, were compared during a 6-month period. In contrast to other studies(14), it was performed largely in an outpa-tient setting.

RESEARCH DESIGN ANDMETHODS

PatientsA total of 95 of the 102 eligible consecutivepatients with NIDDM (mean age [± SD] 68± 9 years, BMl 26.0 ± 4.6 kg'm^, mediantime since diagnosis of diabetes 9 years[range 1-37]; 37 men and 58 women),according to World Health Organization(WHO) ctitena (16), completed the studySeven patients did not complete the studybecause of intercurrent problems, mainlycardiovascular events. All were poorly con-

1326 DlABETtlS CARE, VOLUME 19, NtJMBER 12. DECEMtîER 1996

Wolffenbuttel and Associates

trolled, despite diet and maximal doses oforal anlihyperglycemic agents (15 mgglihenclamide daily, or, in 29 patients,glihenclamide with metformin). Poor hloodglucose control was defined as a fastingblood glucose concentration (mean of lastthree measurements within 2 months)>8.0 mmol/1 and HhAj, >8.0%. A com-plete history and physical examination wasperformed to exclude intercurrent illness orcardiac, hepatic, renal, or other endocrinedisease. No patient had severe untreatedhypertension (diastolic hlood pressure>110 mmHg), impaired kidney function(serum creatinine >140 pmol/1), or wastreated with corticosteroids. All patientsgave their informed consent hefore partici-pating. The study was approved hy the Eth-ical Committee of the University Hospital.

Study designDuring a run-in period of at least 1 month,oral therapy was continued, dietary treat-ment optimized, and instruction in self-monitoring of hlood glucose (if possible)and insulin injections, as well as formaldiabetes education, was given. Thepatients suhsequently started insulin treat-ment with one of the foUovvang schemes:

A. Twice-daily injections of a mixtureof fast-acting and NPH insulin (Mixtard30/70, Novo Nordisk A/S, Bagsvaerd,Denmark, regimen A, n = 34); patientswere instructed to inject the doses ofinsulin suhcutaneously 30 min beforehreakfast and dinner.

B. Combination of NPH insulin (Insu-latard. Novo Nordisk A/S) administeredonce daily at hedtime at 10:00 P.M. (regi-men B, n = 28), with glihenclamide duringthe day (10 mg hefore hreakfast and 5 mghefore dinner). Patients were instructed totake the glihenclamide 5 min heforemeals. When hlood glucose values duringthe day and evening consistently exceeded10.0 mmol/1, patients switched to a twice-daily regimen with insulin hefore hreakfastand the evening meal, while continuingthe glihenclamide.

C. Comhination of NPH insulin (Insu-latard) administered 30 min hefore hreak-fast (regimen C, n = 33), with glihen-clamide during the day (10 mg heforehreakfast and 5 mg hefore dinner). Whenhedtime hlood glucose levels exceeded10.0 mmol/1, a second injection of NPHinsulin hefore dinner was added, withcontinuation of the glihenclamide.

Thus, patients on regimerts B and Ctailing to normalize with one daily injec-

tion converted to the same treatmentscheme, i.e., insulin hefore hreakfast andthe evening meal, in comhination withglihenclamide (regimen D).

To obtain a diurnal profile of glucose,patients were instructed to measure theirblood glucose with an electronic device(Glucometer II, Bayer, Germany, or OneTouch, Lifescan) hefore hreakfast, lunch,and the evening meal, and at hedtime.Insulin dose was adjusted on the basis ofthese hlood glucose measurements. In 26patients, who were for various reasons notcapable or willing to measure their ownhlood glucose, three determinations dur-ing the day, i.e., hefore the main meals,were performed in the hospital 1 weekhefore each visit to the outpatient clinic.Patients not performing home hlood glu-cose measurements were equally distrib-uted among the three groups. Goals oftherapy were fasting hlood glucose levels<7.0 mmol/1, daytime preprandial hloodglucose values <10.0 mmol/1, and HhA] .levels <8.0%. All patients visited the out-patient clinic 2 and 4 weeks after startinginsulin therapy and, thereafter, monthly,when measurements were made of hodyweight, hlood glucose, serum fruc-tosamine, and HhA^.

At baseline and after 6 months, bloodwas drawn in the fasting state for thedetermination of hlood glucose, serumfructosamine, HhA| , nonesterified fattyacids, lipids and apolipoproteins, andserum insulin. Before the initiation of theinsulin treatment, residual ß-cell functionwas assessed hy the response of serum C-peptide and insulin levels to intravenousadministration of 1 mg glucagon.

For comparison, glucagon-stimulatedinsulin and C-peptide secretion and hloodglucose and lipid determinations wereperformed in a group of 18 nonohese con-trol suhjects.

Analytical proceduresBlood glucose was measured in venouswhole hlood vvath an automated hexoki-nase method on the Cohas Bio analyzer.HbAi . was determined with high-perfor-mance liquid chromatography (HPLC)(DIAMAT, Bio-Rad, Richmond, CA; nor-mal values 4.4-6.2%, hetween-assay coef-ficient of variation 2.6%) and serum fruc-tosamine with a Fructosamine Roche kit(Roche, Basel, Switzerland). Cholesterol(CHOD-PAP, Boehringer Mannheim, Ger-many) and triglycérides (TdglyceridRapid, Roche, Basel, Switzerland) were

determined with enzymatic methods on aCohas Bio analyzer. HDL cholesterol wasmeasured enzymatically after precipitationof LDL and VLDL with polyethylene gly-col 6000. Apolipoprotein (apo) Al and Bwere determined v ath an immunoturhidi-metric assay (Roche, Basel, Switzerland)on a Cohas MIRA analyzer. Lipoprotein(a)[Lp(a)l was measured hy radioimmunoas-say (RlA)(Pharmacia, Upp.sala, Sweden).Serum free fatty acids were measured hy acolorimetric method. Seruin free insulinlevels (immunoreactive insitlin) were ana-lyzed with RIA (Pharmacia, Uppsala, Swe-den) after polyethylene glycol pretreat-ment, and serum C-peptide was measuredwith RIA (Mallinckrodt, Germany).Between-assay variation of insulin RIA was7.5% and of C-peptide, 9%.

Statistics and calculationsResidual ß-cell function was assessed inthree ways: J) as the release of C-peptideafter intravenous injection of 1 mgglucagon, calctilated as the incrementalarea under the curve (C-peptide AUC) andthe ratio of C-peptide AUC and fastinghlood glucose (AUCFBG); 2) from the fast-ing levels of hlood glucose, insulin (fastingimmunoreactive insulin), and C-peptide,hy calculating the formulas of Matthews(17); and 3) as the level of C-peptide 5 minafter glucagon injection (C-peptide-„nin).Relative insulin resistance was calculated asR = [fasting serum insulin (mU/1) X hloodglucose]/22.5 (17).

All data are expressed as means ± SDor median and range when not normallydistributed. Calculations were performedwith the SPSS/PC + version for Windows6.0 statistical software package (SPSS,Chicago, IL). Group differences in baselinevariables were compared hy one-wayanalysis of variance for continuous vari-ables and X' tests for categorical variables.Student-Newman-Keuls correction wasapplied for multiple comparisons.Changes of blood glucose control hetweenthe treatment groups (HhA] and fastinghlood glucose) and lipid variables werecompared hy analysis of variance; whenthe changes were not normally distrib-uted, Kruskal-Wallis analy.sis of variancewas used. Linear regression analysis wasused to determine the relationshiphetween two variables. P values <0.05were considered statistically significant.

RESULTS— At haseline, metaboliccontrol, as assessed hy hlood glucose and

DIABETES CARE, VOLUME 19, NUMBER 12, DECEMBER 1996 1327

Insulin therapy in NIDDM

Table 1—Baseline parameters in the various groups and changes of these parameters with insulin therapy on an intention-to-treat basis

Insulin dose (U)Body weight (kg)Fasting blood glucose (mmol/1)Mean blood glucose (mmol/1)Serum fructosamine (pmoUl)Glycated hemoglobin (%)Total cholesterol (mmol/1)HDL cholesterol (mmol/1)Triglycérides (mmol/1)

Apo(Al) (g/1)Apo(B) (g/1)Free fatty acids (mmol/1)Lp(a) (U/1)

Insulin (pmol/1)

Regimen ABaseline

67.4 ± 11.814.5 ± 1.917.1 ±2.6467 ± 5311.2 ±1.37.0 ± 1.2*

1.09 ±0.302.36

(1.11-14.7)1.36 ±0.181.48 ±0.22927 ± 333

236(13-1,381)

64(18-151)

(n = 34)6 months

39 ± 1271.4 ± 12.8t8.5 ±2.0+9.2 ± 1.7t

293 ± 32Î8.2 ± 1.2t6.2 ± L i t

1.21±0.33t1.62t

(0.39-6.6)1.45±0.20t1.31±0.23t581± 297t

264(17-1,420)

81t(30-343)

RegimenBaseline

76.1 ± 16.414.4 ± 1.916.7 ±2.3416 ± 2410.5 ± 1.26.4 ±1.3

1.02 ±0.342.00

(0.82-9.1)1.24 ±0.151.41 ±0.17994 ± 347

281(21-2,470)

71(37-212)

B (n = 28)6 months

24 ±980.5 ± 16.8t8.4 ± 1.9Î9.5 ± 1.2t311 ±21t8.1 ± l . l t6.1±0.9t

1.06 ±0.341.70Î

(0.76-6.6)1.42±0.24t1.32±0.17t470 ± 252t

273(21-3,050)

97t(33-163)

Regimen CBaseline

_

69.2 ± 10.313.6 ±2.717.1 ±3.0433 ± 3611.1 ± 1.36.1 ± 1.2

1.05 ±0.322.21

(1.03-7.0)1.36 ±0.191.35 ±0.28

1,001 ±338170

(8-1,142)67

(34-286)

(n = 33) Nondiabetic6 months control subjects

26 ± 1772.6 ± 10.7t8.1±2.2t9.9±1.9t

320±.25T8.5±l . l t5.8 ± l.Ot

1.10±0.36t1.91t

(0.64-4.4)1.39 ±0.221.28±0.24t607 ± 304t

206(8-1,148)

92t(33-360)

4.1 ±0.34.5 ±0.4

5.2 ±0.55.0 ±0.4

1.40 ±0.291.50

(0.51-2.00)1.53 ±0.221.13±0.15480±120

101(8-1,747)

Dala are means ± SD or median (range). 'P < 0.05 vs. regimen C; tP < 0.05 vs. baseline (corrected for multiple comparisons).

ii^ levels, was poor (Table 1). No dif-ferences in glycémie control and serumlipids was observed between the random-ization groups, besides slightly higherserum cholesterol in group A patients.Residual ß-cell function was slightly betterin patients starting with regimen C, espe-cially, the increment of G-peptide levelsafter glucagon levels, either calculated asarea under the curve (G-peptide AUG) orcorrected for fasting blood glucose(AUGEBG), was higher in this group(Table 2). HbAi at the start of the studywas inversely correlated with insulinsecretion (HbA,, vs. AUGEBG, r = -0.45,P < 0.001; HbA|, vs. G-peptide AUG, r =-0.33, P = 0.002) but not with calculatedinsulin resistance. This suggests that thepoor glycémie control was caused mainlyby ß-cell insufficiency.

During insulin therapy, blood glucosecontrol improved considerably In the totalpatient population, HbAij ^ levels decreasedfrom 11.0 ± 1.3 to 8.3 ± 1.2% (P < 0.001),and fasting blood glucose levels decreasedfrom 14.1 ± 2.2 to 8.3 ± 2.0 mmol/1 (P <0.001). Mean increase in body weight was4.0 kg. Also, significant reductions wereobserved in total cholesterol, triglycérides,apoB, and free fatty acids, whereas HDLcholesterol and apoAl increased by 10%.

Table 1 and Eig. 1 depict the changesin the three randomization groups on thebasis of an intention-to-treat analysis.Insulin dose was significantly lower inpatients on combination therapy (regi-mens B and G). The decrease of fastingand daily mean blood glucose, and HbAiwas not different in the three groups, nei-ther was the increase in body weight. Of

the patients, 34 achieved HbAi levels<8.0% (15 of 34 on regimen A, 11 on reg-imen B, and 8 on regimen G); 25 patientseven achieved HbAi levels <7.5%. Sub-jects stariing with two injections weremore likely to achieve good glycémie con-trol than subjects starting with one injec-tion (Table 3). The improvement of meta-bolic control, calculated as the decrease ofHbAu between baseline and 6 months,correlated with baseline HbAi (r = 0.63, P< 0.001) and with AUGEBG (r = -0.39,P = 0.001), but not with insulin dose.

Similar improvements of lipid abnor-malities were observed, although thereduction of serum cholesterol was slightlygreater in group A patients, probablybecause of the higher cholesterol levels atbaseline. The largest reduction of free fattyacid levels was observed in group B (bed-

Table 2—Baseline values of indices of ß-cell function and insulin resistance in the various groups

Regimen A Regimen B Regimen CNondiabetic

control subjects

I asting C-peptide (pmol/1)Fasting insulin (pmol/1)C-peptide,„„„ (pmol/1)C-peptide AUC (nmol/1 • 30 min)AUCFBGß-cell function-C-peptide (%)ß-cell function-insulin (%)Relative insulin resistance

730 ±41760(18-151)1,081 ± 5335.7(1.9-18.6)

381 (153-1,358)16 (4-61)15(5-42)

5.5(2.5-13.0)

887 ± 42471 (37-212)1,340 ±6185.4(0.7-18.6)346(54-1,721)

22 (9-85)17 (9-59)

5.8(2.6-21.9)

913±41567 (34-286)1,503 ±774

9.1(1.0-32.6)*708(64-2,217)*23 (10-499)20(10-589)5.2(2.1-26.3)

551 ±20146 (32-66)1,760 ±400

29.8(19.8-45.1)7,345 (5,016-10,620)

150(100-312)120(84-159)1.4(1.0-2.3)

Data are means ± SD or median (range). *P < 0.05 vs. regimens A and B.

1328 DiABErES CARE, VOLUME 19, NUMBER 12, DECEMBER 1996

Wolffenbuttel and Associates

o

12.0 -,

11.0 -

10.0 \

9.0

8.0

7.0

Two-injections (A)Evening insulin + S.U. (B)Morning insulin + S.U. (C)

\

2 3 4

Time (months)

Figure 1—Changes during Insulin therapy oj HbA¡^ in the three different treatment regimens. A-Crepresent the different insulin regimens on an intention-to-treat basis. Regimen A: twice-daily insulininjections; regimen B; evening insulin plus daytime sulfonylurea; regimen C: morning insulin plus day-time sulfonylurea.

time insulin), altbougb the difference witbtbe otber groups was not statisticallysignificant. Baseline Lp(a) levels were ele-vated compared witb nondiabetic controlsubjects, but with insulin therapy, nochanges of serum levels of Lp(a) wereobserved.

Nine patients who were starting withthe once-daily insulin regimen B and 11patients starting with regimen C needed asecond insulin injection. The insulin doseof these patients after 6 months of treat-ment (i.e., regimen D) was significantlyhigher (34 ± 19 U) than that of tbepatients wbo could be maintained on oneinjection (23 ± 9 and 20 ± 7 U, respec-

Table 3—Number of patients achievinggood giycemic control or remaining insuffi-ciently controlled after 6 months

tively). But eventually, comparableimprovements of metabolic control (Eig.2) and serum lipids were acbieved. Sub-jects who needed a second insulin injec-

I I Two-injections (A)^ ^ Evening insulin * SU, (B)^ ^ Morning insulin + S.U. (C)

I Two-injections + SU. {D)

Ç. 0

á- -6 -

Tviiice-daily insulinBedtime insulin

plus sulfonylureaMorning insulin

plus sulfonylureaTotal

*X significance 0.03.

<8.0%

1511

8

34

>9.0%

66

13*

25

-10

I

tion to improve metabolic control badlower levels of fasting C-peptide andinsulin, as well as C-peptide levels 5 minafter intravenous glucagon. The other ß-cell parameters were slightly, but notsignificantly lower (Table 4).

One patient wbo was treated with reg-imen C suffered from severe hypo-glycemia, necessitating help from othersby injection of glucose. This occurred dur-ing a gastrointestinal infection. He recov-ered completely No other patients suf-fered from severe hypoglycemia. Mostpatients reported improvement of well-being and (if present before institution ofinsulin therapy) abolishment of byper-glycemic complaints.

CONCLUSIONS— Secondary failureto oral antihyperglycemic agents after aninitial favorable response may occur in—5-10% of patients with NIDDM on ayearly basis (18,19). The results of thisstudy confirm the effectiveness of insulintherapy, alone or in combination withglibenclamide, to improve blood glucosecontrol. In all treatment groups, a cleardecrease of fasting and mean daily bloodglucose and a 25-30% decrease of HbA,cwas observed. Mean HbA], levels of 8.3%were achieved, which is intermediatebetween the conventional and intensive

A B C D

Figure 2—Changes after 6 months of insulin therapy of body weight and measures of glycémie con-trol. A-D represent the different insulin regimens on a post hoc analysis basis. Regimen A: twice-dailyinsulin injections (n = 34); reamen B: subjects who remained on one evening insulin injection plusdaytime sulfonylurea throughout the study (n = 19); reamen C: subjects who remained on one mom-ing insulin injection plus daytime sulfonylurea throughout the study (n = 22); regimen D; subjects whostarted with one daily insulin injection but needed a second insulin administration (n = 20).

DIABETES CARE, VOLUME 19, NUMBER 12, DECEMBER 1996 1329

Insulin therapy in NIDDM

treatment groups of the Diabetes Controland Complications Trial (20). It must betaken into consideration that our patientswere considerably older, with an averageage of almost 70 years. Patients using twiee-daily insulin injections were more likely toachteve HbAi,' levels <8.0% (Table 3).However, insulin dose was also the highestin these subjects, compared with those whoreceived combined treatment. In addition,it may be postulated that postprandialblood glucose levels were lower in thesesubjects because they received an insulinmixture that included fast-acting insulin.There was no difference in the increase ofbody weight between the three treatmentgroups. Of the patients who were initiallyrandomized for once-daily insulin injec-tions in combination with glibenclamide,20 (33%) did not reach adequate improve-ment of metabolic eontrol and needed asecond insulin injection.

Approximately 40 years ago, it was rec-ommended that insulin be combined withsulfonylureas in the treatment of NIDDMsubjects poorly controlled with tabletsalone (21). Thereafter, a period followed inwhich the sole use of higher doses ofitisulin was recommended rather than thecombination with sulfonylureas. Theresults of the widely criticized UniversityGroup Diabetes Project (UGDP) study,which suggested an increased risk of car-diovascular deaths in patients treated withsulfonylureas (22), also contributed to thisview. The last 10-15 years, however, showa revival of combined therapy (23-25).Most studies reported during short-termobservations that patients treated with thecombination of insulin and sulfonylurearequired a lower insulin dose and exhibitedbetter glycémie control than patients usinginsulin and placebo. Despite these favorableeffects of combined therapy, it was, in gen-eral, difficult to attain (near) normo-glycemia in the individual patient. Somestudies have reported only transient benefi-cial effects of combined therapy (24,25).

Riddle and colleagues (12,13) havewidely advocated the use of a regimencombining bedtime NPH insulin with day-time sulfonylureas. This approach aims atreducing fasting blood glucose values bysuppression of the increased (nocturnal)hepatic glucose production and high noc-turnal concentrations of free fatty acids,which may contribute to hepatic insensi-tivity to insulin and fasting hyperglycemia.It was suggested that this approach wouldprevent an increase of body weight. Taski-

Table 4—Differences of ß-cell function and insulin resistance in patients who remained onone insulin injection and those who needed a second insulin injection to improve

One injection Two injections

Fasting C-peptide (ptnol/1)Fasting insulin (pmol/1)C-peptide5n,,n (pmol/1)C-peptide AUC (nmol/1 • 30 min)AUCFBGß-cell function C-peptide (%)ß-cell function insulin (%)Relative insulin resistance

922 (300-2,350)76 (39-286)

1,405(640-4,260)7.4(1.85-32.6)

532 (136-2,217)24 (10-499)20(11-589)

6.0 (2.7-26.3)

752(330-1,295)*62 (34-149)*

1,115(450-2,010)*6.6(0.71-15.5)428(54-1,385)

20 (9-51)17 (9-40)

4.6(2.1-12.9)**P < 0.05 vs. one injection.

nen et al. (26) raised the possibility that theimprovement in glycémie control by bed-time insulin was related to the reduction ofplasma free fatty aeids concentrations. Thishypothesis remains to be proven. Our datashow a slightly, but not significantly (P =0.18), larger reduction of fasting plasmafree fatty acids in the bedtime insulingroup; however, the decrease of HbAi wasnot greater in this group compared withthe other regimens.

In a prospective study of 3 months'duration in tablet-treated NIDDMpatients, Yki-Jarvinen et al. (14) showedthat the addition of NPH-insulin in theevening equally improved metabolic con-trol as did a two-injeetion or a multiple-injection regimen. However, weight gainwith evening insulin was less: a meanincrease of 1.2 kg was observed in thisgroup, against 1.8 kg and 2.9 kg in theother groups. We did not observe differ-ences in weight gain between the differentinsulin regimens. But, metabolic controlbefore institution of insulin treatment wasconsiderably poorer in our patients com-pared with Yki-Jarvinen's patients. At thestart of the study, mean HbA,c, measuredwith comparable methodology and refer-ence values, was 11.0% in our patients vs.9.7% in theirs, while after 6 and 3 monthsof insulin therapy, respectively, these levelswere 8.3 vs. 8.1%. These results are inagreement with the observations that thechange in body weight was inversely cor-related with the ehange in glycated hemo-globin. With regard to serum insulin lev-els, the average increase of fasting insulinin our patients was 40% and was compa-rable between the treatment regimens. Wedid not perform 24-h insulin profiles,which were shown to increase less withcombined therapy than with insulin treat-ment alone (14). In addition to the effeets

on glycémie control, we observed a signifi-cant decrease of total serum cholesteroland triglycérides, both in patients oninsulin alone and in patients on combinedtherapy. Also, an increase of HDL choles-terol was observed. Thus, with insulintherapy, a less atherogenic profile wasachieved. In aceordance with earlierreports from our group (27), we did notobserve any changes of serum Lp(a) levels.Total cholesterol and triglycéride levels,however, remained elevated, and HDLcholesterol remained somewhat lowercompared with nondiabetic subjects(Table 1), as was already demonstrated byothers (6,7). Concomitant with theincrease of HDL cholesterol, we observedan increase of apoAl, suggesting anincrease in the total number of HDL parti-cles. Our findings are in contrast withobservations made by Stenman et al. (15),who reported that during combined ther-apy, HDL cholesterol was significantlylower than with insulin alone. Theseobservations were made in a small num-ber of only 15 patients who were treatedwith insulin alone (most of them only oneinjection) or insulin in combination withglibenclamide in a 4-month crossoverstudy.

Groop et al. (19) studied possible fac-tors influencing secondary drug failureduring treatment with oral antihyper-glycemic agents. Patients failing to sulfony-lurea had a lower relative body weight andlower basal and post-glucagon plasma C-peptide concentrations than patients whowere well controlled. In a previous study,we measured C-peptide levels before andduring 15 min after intravenous injectionof glueagon in NIDDM patients failing ondietary therapy (28). An incremental areaunder the response curve of <5.0 nmol/1 •15 min identified those patients who

1330 CARE, VOLUME 19, NUMBER 12, DECEMBER 1996

Wolffenhuítel and Associates

Table 5—Estimated monthly costs of insulin and combined insulin-sul/onylurea therapy inpatients with NIDDM who inject themselves with use of an insulin pen

Number of insulin injectionsAverage amount of insulin (U/day)Costs (Dfl.)

InsulinNeedlesGlihenclamide ( 15 mg/day)Insulin pen

Total costs (Dn.)

Insulin alone

239

70.1224.270.00

Free of charge94.39

(U.S. $59.00)

Insulin plus sulfonylureas

l o r 225

44.9518.2048.70

111.85(U.S. $70.00)

. Dutch guilders. 1 U.S. $ = Dil. 1.60.

exhibited better glycémie control oninsulin therapy compared with treatmentwith sulfonylureas. The finding in the pres-ent study that HbAi . was inversely corre-lated with glucagon-stimulated insulinsecretion implies that secondary failure tooral agents was mainly caused by ß-cellinsufficiency However, this insufficiencywas not adequately reflected by other vari-ables, such as fasting insulin or C-peptidelevels, or by calculation of specific ß-cellfunction formulas, which make use ofthese parameters. It must be noted in thisrespect that combining sulfonylureas andinsulin was successful only in thosepatients, who exhibited residual ß-cellfunction, as estimated by fasting and post-glucagon C-peptide levels (29). Also in ourpatients, those who needed a secondinsulin injection showed lower fastingplasma C-peptide and insulin levels and C-peptide response after glucagon, thus hav-ing lower residual islet ß-cell functionbefore the initiation of insulin treatment.However, glucagon-stimulated insulinsecretion, although it followed the samepattern, was not related. Thus, parametersof insulin secretion cannot be used to pre-dict in the individual patient the efficacy ofinsulin therapy

Results of long-term studies in whichdifferent insulin regimens have been com-pared are scarce. Some authors advocatethat patients start with the least complextreatment, for instance continuation oftheir tablet treatment and the addition ofinsulin at bedtime. Because maximal oraltherapy often comprises the use of twodrugs (i.e., a sulfonylurea combined withmetformin, as 60% of Yki-Jar\'inen'spatients were using), one may wonderwhether introducing a third medicationfor the control of diabetes is more benefi-

cial for the patient than completely switch-ing to insulin therapy alone. Based on thepresent results, the latter approach seemsto be more realistic in light of other treat-ments that patients may be receiving forhypertension, dyslipidemia, cardiovascu-lar disease, etc. An additional disadvantagemay be the higher cost of combined ther-apy Table 5 gives an estimate of the cost oftreatment of the individual patient, asderived from the use of insulin and gliben-clamide in the different patient groups ofour study Combined therapy was —15%more expensive than insulin alone.

We conclude that in elderly patientswith NIDDM, improvement of glycémiecontrol can be achieved at the expense ofsome weight gain. Insulin treatment hasmoderate beneficial effects on serumlipoproteins. Although on the basis ofglycémie control and weight gain, no pref-erence for any treatment regimen can bediscerned, twice-daily insuhn administra-tion is the most simple and cost-effectiveregimen.

Acknowledgments— This study was sup-ported by a grant from the Diabetes ResearchFoundation (Diabetes Fonds Nederland).

References

1. SkylerJS: Non-insulin-dependent diabetesmellitus: a clinical strategy. Diabetes Care 7(Suppl. 1): 118-129, 1984

2. Gerich JE: Drug therapy: oral hypo-glycémie agents. N Eng! J Med321:1231-1245,1989

3. Tattersall RB, Scott AR: When to useinsulin in the maturity onset diabetic. Post-grad MedJ 63:859-864, 1987

4. Genuth S: Management of the adult onsetdiabetic with sulfonylurea drug failure. In

Endocrinology and Metabolism Clinics ofNorih America. Diabetes Mellitus—Perspec-tives on Therapy. Vol. 21. Karam JH, Ed.Philadelphia, Saunders, 1992, p. 351-370

5. Wolffenbuttel BHR, Weher RFA, VanKoetsveld PM, Weeks L, Verschoor L: Arandomized crossover study of sulfony-lurea and insulin treatment in patientswith type 2 diahetes poorly controlled ondietai-y therapy Diabetic Med 6:520-525,1989

6. Lindstrom T, Arnqvist HJ, Olsson AG:Effect of different insulin regimens onplasma lipoprotein and apolipoproteinconcentrations in patients with non-insulin-dependent diabetes mellittts. Alh-nmlerosis 81:137-144, 1990

7. Bagdade JD, Buchanan WE. Kuusi T,Taskinen M-R: Persistent abnomialities inlipoprotein composition in noninsulin-dependent diahetes after intensive insulintherapy ArieriosclcTtKis 10:232-239, 1990

8. Montgomery BJ: High plasma insulin levela prime risk factor for hcail disease. JAMA241:1665, 1979

9. Stout RW: The role of insulin in athero-sclerosis in diabetics and nondiabetics.Diabetes 30 (Suppl. 2):54-57, 1981

10. Nilsson A, Tideholm B, Kalen J, KatzmanP; Incidence of severe hypoglycemia andits causes in insulin-treated diabetics. AdaMed ScancJ 224:257-262, 1988

11. Wolffenbuttel BHR: Combined use ofinsulin and oral hypoglycaemic agents inthe treatment of patients with diabetesmellitus. Eurjlnt Med 1:89-95. 1989

12. Riddle MC, HartJS, Bouma Dj, PhillipsonBE, Youker G: Efficacy of hedtime NPHinsulin with daytime sulfonylurea for sub-population of type 11 diabetic subjects.Diabfifs Care 12:623-629, 1989

13. Riddle MC: Evening insulin sti'atcgy Piti-bftcsCüre 13:676-686, 1990

14. Yki-Jar\'inen H, Kauppila M, Kujansuu E,LahtiJ, Matjanen T, Niskanen L, R:\jala S,Ryysy L, Salo S, Seppala P: Comparison ofinsulin regimens in patients with non-insulin-dependent diahetes mellitus. NEngl;Mc-d 327:1426-1433. 1992

15. Stenman S, Groop PH. Saloranta C, Totter-man KJ, Fyhrqvist ¥. Groop L: Effects ofthe combination of insulin and gliben-clamide in type 2 (non-insulin-dependcnt)diabetic patients with secondary failure tooral hypoglycaemic agents. Diahetologia31:206-213, 1988

16. World Health Organization: WHO ExpertCommitlee on Diabetes MdlUus. SecondReport. Geneva, World Health Org., 1985(Tech. Rep. Ser. no. 727)

17. Matthews DR, Hosker jP Rudenski AS,Naylor BA, Treacher DF, Turner RC:Homeostasis model assessment: insulinresistance and ß-cell function from fastingplasma glucose and insulin concentrationsin man. Diabetobgia 28:412-419, 1985

DIABETES CARE, VOLUME 19, NUMBER 12, DECEMBER 1996 1331

Insulin therapy in NIDDM

18. Thoelke H, Ratzmann KP: Häufigkeit desSekundärversagens einer Sulfonylham-stofftherapie-Epidemiologische Unter-suchung. Dtsch Med Wschr 114:580-583,1989

19. Groop LC, Pelkonen R, Koskimies S, Bot-tazzo GF; Doniach D: Secondary failure totreatment with oral antidiabetic agents innon-insulin-dependent diabetes. DiabetesCore 9:129-133, 1986

20. The Diabetes Control and ComplicationsTrial Research Group: The effect of inten-sive treatment of diabetes on the develop-ment and progression of long-term com-plications in insulin-dependent diabetesmellitus. NEngiJMecl 329:977-986, 1993

21. Lazarus SS, Volk BB: Physiological basis ofthe effectiveness of combined insulin-tolbutamide therapy in stabile diabetes.Ann N YAcadSd 82:590-602, 1959

22. Feinstein AR: The persistent clinical fail-ures and fallacies of the UGDP study C!in

Pharmacol Therap 19:78-93, 197623. Hamelbeck H, Klein W, Zoltobrocki M,

Schofñing K: Glibenclamide-insulin com-bination in the management of secondaryfailure of sulfonylurea medication. DtschMed Wschr 107:1581-1583, 1982

24. Osei K, O'Dorisio TM, Falko JM: Con-comitant insulin and sulfonylurea therapyin patients with type II diabetes: effects onglucoregulation and lipid metabolism. AmJ Med 77:1002-1009, 1984

25. Groop L, Hamo K, Nikkila EA, PelkonenR, Tolppanen EM: Transient effect of thecombination of insulin and sulfonylurea(glibenclamlde) on glycémie control innon-insulin-dependent diabetics poorlycontrolled with insulin alone. Acta MedScand 217:33-39, 1985

26. Taskinen MR, Sane T, Helve E, KaronenSL, Nikkila EA, Yki-Jarvinen H: Bedtimeinsulin for suppression of ovemight free-fatty acid, blood glucose, and glucose pro-

duction m NIDDM. Diabetes 38:580-588,1989

27. Wolffenbuttel BHR, Leurs PB, Sels JPJE,Rondas-Colbers GJWM, Menheere PPCA,Nieuwenhuijzen Kruseman AC: Improvedblood glucose control by insulin therapyin type 2 diabetic patients has no effect onlipoprotein(a) levels. Diabetic Med10:427-430, 1993

28. Wolffenbuttel BHR, Weber REA, VanKoetsveld PM, Weeks L, Verschoor L: Arandomized crossover study of sulfonylureaand insulin treatment in patients with type2 diabetes poorly controlled on dietarytherapy E)iabe(!c Med 6:520-525, 1989

29. Castillo M, Scheen AJ, Paolisso G, LefebvrePJ: The addition of glipizide to insulintherapy in type 11 diabetic patients withsecondary failure to sulfonylureas is usefulonly in the presence of a significant resid-ual insulin secretion. Acta Endocrinol116:364-372, 1987

1332 DIABETES CARE, VOLUME 19, NUMBER 12, DECEMBER 1996

Copyright of Diabetes Care is the property of American Diabetes Association and its content may not be copied

or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission.

However, users may print, download, or email articles for individual use.