continuous Subcutaneous Insulin Infusion in the ... - Diabetes Care

2 downloads 0 Views 2MB Size Report
in the Treatment of Diabetes Mellitus. J. C. PICKUP, H. KEEN, G. C. VIBERTI, M. C. WHITE, E. M. KOHNER, J. A. PARSONS,. AND K. G. M. M. ALBERTI.
c

ontinuous Subcutaneous Insulin Infusion in the Treatment of Diabetes Mellitus J. C. PICKUP, H. KEEN, G. C. VIBERTI, M. C. WHITE, E. M. KOHNER, J. A. PARSONS, AND K. G. M. M. ALBERTI

Low-volume, dual-rate, continuous subcutaneous insulin infusion (CSII) creates long periods of nearnormalization of blood glucose and major intermediary metabolites in most insulin-requiring diabetic patients. The technology and strategy of the system are discussed. We have observed encouraging clinical and fluorescein angiographic improvement in severe diabetic retinopathy after 3 mo of outpatient CSII; in the kidney, glomerular capillary permeability (microalbuminuria) is reduced or normalized in long-standing diabetic patients after a few days of CSII-induced strict control. Reduction in insulin dose during CSII treatment of newly diagnosed ketonuric diabetic patients may indicate improved B-cell function in this group. Although CSII must remain a research tool, undertaken only under close medical supervision, it is increasingly likely that the technique affords the conditions for testing the hypothesis that metabolic near-normalization of diabetes slows, arrests, or reverses the course of the microvascular disease associated with the syndrome, DIABETES CARE 3.- 290-300, MARCH-APRIL 1980.

W

ith the advent of continuous blood glucose monitoring1*2 and frequent sampling profiles it has become apparent that in most insulin-dependent diabetic patients it is very difficult to sustain near-normal blood glucose concentrations. The increasing evidence from animal studies for the importance of good diabetic control3*4 has, therefore, stimulated the search for new ways of delivering insulin and improving metabolic control in diabetic patients. The three most promising approaches are probably (1) islet tissue transplantation,5 (2) self-monitoring of capillary blood glucose samples6*7 with glucose oxidase reagent strips and frequent corrective adjustments of insulin dosage, and (3) electromechanical devices for infusing insulin in a more physiologic way than is possible with conventional injections.8 We have concentrated on the last option, although our own infusion system also makes use of the advantages of glycemic self-monitoring. The idea of delivering insulin by continuous infusion dates back to 1934.9 Metcalf, a medical student at the London Hospital, demonstrated the blood-sugar-lowering effects of both intravenous and subcutaneously infused insulin in a small group of diabetic patients over a period of a few hours. The rediscovery10 and development11 of the technique of continuous subcutaneous insulin infusion (CSII), more than

290

40 yr later, grew out of one specific aim: to find a method to test the hypothesis in man that the evolution of early diabetic microangiopathy responds to the metabolic normalization of the disease. Replacement therapy in endocrine deficiency diseases should aim to mimic blood levels of hormone that occur in the normal person. In nondiabetic humans, insulin is delivered at essentially two rates: a slow, basal rate during the night and between meals, which suppresses and controls glucose output from the liver, and an augmented prandial secretion, which disposes of absorbed nutrients. The basal insulin supply is particularly difficult to reproduce in diabetic patients because the short half-life of insulin demands either a long-acting preparation or continuous delivery. The former often fall short of the ideal for a variety of reasons, e.g., unpredictable absorption with varying and often inappropriate duration of action. The so-called artificial endocrine pancreas (AEP)12*13 is a device for continuous administration of insulin that incorporates blood glucose sensing, which feeds back to control the infusion rate ("closed-loop"). Although very successful in maintaining normoglycemia in diabetic patients for periods of up to a few days, it has major disadvantages for long-term use, and it is hardly conceivable that, in its present form, it could be used to achieve the sufficiently prolonged strict con-

DIABETES CARE, VOL. 3 NO. 2, MARCH-APRIL 1980

CONTINUOUS SUBCUTANEOUS INSULIN INFUSION/J. C. PICKUP AND ASSOCIATES

TABLE 1 Requirements for devices intended to demonstrate the effect of control on diabetic complications 1. Control must be long term. 2. Good control of not only blood glucose but also intermediary metabolites (and possibly other hormones such as glucagon). 3. Must be safe, cheap, reliable. 4- Must be portable, so that patient activity is unrestricted. 5. Delivery rate must be controllable, i.e., at least dual rate. 6. Delivery point should be complication free. 7. System should be suitable for most types of diabetes (to enable random allocation to a trial group).

trol required to demonstrably alter the course of diabetic microangiopathy. The AEP is, at present, a costly and bulky bed-side apparatus and complex enough to demand constant attendance by a team of doctors and technicians. Furthermore, insulin is infused into a peripheral vein, a route potentially associated with thrombosis, phlebitis, septacemia, and infective thromboembolism. Slama et al.14 showed in 1974 that control almost as good as with AEP can be achieved with a simple, portable intravenous infusion system without a glucose sensor ("open-loop") but employing just two rates of insulin delivery. There are now many such systems,15"20 some of which have been used to treat a small number of diabetic patients for up to 3 mo under outpatient conditions.20 Nevertheless, the aforementioned hazards of prolonged intravenous systems attracted us to continuous subcutaneous insulin infusion as a possible way of approaching the ideal requirements for a portable and stable delivery system (outlined in Table 1). TECHNIQUES

Pump technology. Our prototype infuser was a specially constructed, battery-operated syringe pump developed at the

National Institute for Medical Research, Mill Hill, London.21 In this model a quartz crystal oscillator and integrated circuit divider provide gating pulses every few seconds to energize a small electric motor. The drive is geared to a lead screw, which advances, via a movable lever, the plunger of an ordinary disposable plastic syringe (Figure 1A). Originally the pump was electronically adapted to deliver at two rates: 50 fiUh. (basal) and 400 fxUh (meal-time). The patient pressed a guarded button on the end of the pump to engage the high rate, which returned automatically to 50 (JLU\\ after 17 min. In the current production model (marketed by Muirhead Ltd., Beckenham, Kent, U. K.), the lead screw can be turned manually and the syringe plunger advanced by a knob on the outside of the pump, thus giving at meals a measured insulin bolus (as opposed to a short infusion), the use of which can be selected by the number of countable turns of the screw. As described below (pharmacokinetics section) there is now evidence that the latter mechanism is at least equally effective for diabetes and has the merit of simplicity and flexibility, allowing an infinite ratio of basal to prandial insulin supply. The pump operates efficiently against back pressures of at least 50 psi. The mercury battery has an approximately 2-wk life but is usually changed weekly for safety. The size of the present infuser is 144 X 70 X 23 mm, and the weight is about 300 g. The Pye Dynamics Syringe Driver (Pye Dynamics Ltd., Bushey, Hertfordshire, U. K.) and the Autosyringe (Autosyringe Inc., Hooksett, New Hampshire) have been successfully used for CSII by other groups. The Autosyringe22 is larger and heavier than the Muirhead model but has a flexible rate adjustment. The Pye infuser23 accepts a number of syringe sizes and is similar in size to the Muirhead but requires the patient himself to select the meal rate and later adjust it to basal. Our experience of long-term outpatient use of pumps indicates that the ideal version for clinical use should be small, light, reliable, mechanically and electroni-

FIG. I (A) Inside of the Mill Hill Infuser showing disposable plastic syringe. (B) Infuser fitted to a patient. The distal end of the nylon cannula is subcutaneously implanted in the abdominal wall.

DIABETES CARE, VOL. 3 NO. 2, MARCH-APRIL 1980

291

CONTINUOUS SUBCUTANEOUS INSULIN INFUSION/J. C. PICKUP AND ASSOCIATES

cally simple and easy for the patient to operate. These demands have not yet been fully met. Infusion cannulae and implantation site. Most of our stud-

ies24"27 employ a fine nylon cannula (internal diameter 0.2 mm) manufactured by Portex Ltd., Hythe, Kent, U. K. The distal 3 cm or so are implanted in the subcutaneous tissue of the abdomen.24 Before insertion, the skin is cleaned with antiseptic solution and then anesthetized in two small areas about 6 cm apart using 1% lignocaine solution. An introducing needle (e.g., 16-guage "Medicut" with plastic sheath, Sherwood Medical Industries) is inserted through one anesthetized area and pushed through subcutaneously to emerge through the other. An appropriate length of the nylon cannula is fed into the lumen of the needle via the tip. Withdrawal of the needle now leaves the cannula subcutaneously implanted, and it is secured with adhesive tape. An alternative cannula is a fine guage metal needle attached to connecting tubing (e.g., "Butterfly," Abbott Laboratories, Queenborough, Kent, U. K.) or Deseret22 (Deseret Company, Sandy, Utah). These needles can be easily implanted by the patient himself and so are useful as an emergency replacement if the nylon cannula is inadvertently withdrawn. They probably need to be reimplanted at a separate site every day or so to avoid irritation. By comparison, the much finer nylon cannula has been left at a single site for 2 mo and is on average changed every 2 - 3 wk. The small diameter minimizes release of insulin from the tip caused by random flexing of the cannula. We have never seen infection at the implantation site nor does significant lipodystrophy occur. The absence of tissue reaction is probably because of the use of highly purified porcine insulin in our infusion studies to date. We have had reports that traditional soluble bovine insulin does cause such changes. The choice of the abdomen for implantation was based on a number of considerations. Firstly, the pump is conveniently worn nearby, around the waist or in a shoulder holster. Binder28 has also shown that subcutaneous (depot) injections of radioiodinated insulin is better absorbed from the abdomen than the limbs, and one would expect less exercise-induced hypoglycemia than if insulin were to be delivered to the subcutaneous tissue of the legs (see below). Nevertheless, CSII has been successfully undertaken using the thigh area, without problems of hypoglycemia (A. Schiffrin, personal communication). Infusion strategy. The highly purified insulin is diluted according to each patient's dosage requirements with sterile 0.154 M saline. Patients replace the insulin in the pump syringe daily, and this is probably important in avoiding precipitation of insulin in the cannula or reservoir, a problem with many intravenons infusion systems in which insulin is kept at near-body temperature in a reservoir for many days.20'29 We have had no proven episode of cannula blockage during CSII. The high rate is usually engaged or the bolus boost delivered 30 min before each main meal to allow for insulin absorption. Champion et al.30 have advocated 15 min before

meals, but this does not agree with our own clinical experience in most patients or with our pharmacokinetic data. Naturally, though, the absorption characteristics will vary somewhat from patient to patient, just as is the case with injection therapy. Total insulin dosage used in the patient's conventional regime is used as the basis for setting the initial insulin infusion rate.25 Usually we start with 100% of the normal total daily dose, including meal-time step-ups; depending on response, this total is increased or decreased in 10—20% steps until near-normogrycemia is achieved. With the electronic 8-fold step-up rate increase, this strategy resulted in about 70% of the daily dose being given at the basal rate and final stabilized basal rates of about 1.3-1.8 U/h with the rest in the three pre-meal boosts. More recently, with the bolus boost model, only 30-40% of the total conventional dose (about 45% of final CSII dose) has been given at the basal level, about 0.75-1 U/h. Tamborlane et al.22 have reported good control with 12.5-15 mU/kg/h, and Champion et al.30 with a basal rate of 30% of total prevalent dose. The size of the prandial increase with the bolus boost model can be varied; often about 5 0 - 60% of total CSII dose is divided between each main meal, more insulin being needed before breakfast than other meals. In our study of six diabetic patients treated by CSII at home for up to 4 mo,27 patients were taught to monitor their own capillary blood glucose concentration with Dextrostix (Ames-Miles) and a reflectance meter. If the glucose concentration rose to 11-15 mmol/L they were instructed as immediate action to give an extra 17-min insulin boost, and if values were above 15 mmol/L to inject a prestated insulin dose subcutaneously by conventional means. For bathing or other purposes, the pump can be disconnected from the cannula without loss of metabolic control for about 20 min, the hub being protected with a cap or spiggot. All of our patients treated at home have been given continuous access to medical advice and assistance. This standard of care is mandatory.

CONTROL OF GLYCEMIA

T

292

argets. In the search for improved degrees of control, using new devices, what are the targets? Optimally we would wish to achieve continuous physiologic normality, but, accepting that this is not yet entirely feasible, to what degree can we allow deviation from it? Symptomatic hypoglycemia is a strikingly uncommon problem with CSII, perhaps for the reasons that are discussed later. Does biochemical (though asymptomatic) hypoglycemia carry risks to health, either short- or long-term? A theoretical case can be made for attempting to keep blood glucose concentrations above 3 mmol/L, allowing an occasional dip into a symptomatic or mildly symptomatic hypoglycemia, which, if persistent, can be corrected. What upper level should be set in attempting to achieve the aim of pre-

DIABETES CARE, VOL. 3 NO. 2, MARCH-APRIL 1980

CONTINUOUS SUBCUTANEOUS INSULIN INFUSION/J. C. PICKUP AND ASSOCIATES

vention of microangiopathy? And, in this phase of our studies, what degree of control must we achieve to assert that experimental failure to demonstrate an effect was not due to inadequate control? It is clear that anything short of normal could be held to explain a failure to reverse microangiopathy, but there are hints that glycemic levels below a certain "threshold" may be adequate to demonstrate an effect. Thus, with comparatively little demonstrable change in blood or urine glucose, Job et al.30 appeared to show a beneficial effect of control on retinopathy. In 10 yr of follow-up of individuals in Bedford and Whitehall with "borderline diabetes" (i.e., a degree of glucose intolerance characterized by blood glucose concentrations between 120 and 200 mg/dl 2 h after an oral 50 g glucose load), significant ophthalmoscopic evidence of retinopathy was not seen.32 By contrast, people with 2-h blood glucose concentrations exceeding 200 mg/dl were clearly susceptible to development of retinopathy.32 A similar threshold effect for retinopathy and nephropathy was seen in the Pima Indians.33'34 However, this should not be understood as indicating that blood glucose below 200 mg/dl is harmless. That level is only an index, standardized, as it were, by the glucose load; blood glucose concentrations in borderline diabetic patients during an ordinary day with regular meals very rarely rise above 150 mg/dl (8.5 mmol/L). We have not completely attained but seek to achieve the range 2.5 to 8.5 mmol/L as the extreme limits of glycemia. It seems unlikely that such additional measures as glycosylated hemoglobin estimation can do more than provide a confirmatory blanket check useful perhaps for screening out definite failures of glycemic control. Results. In fully ambulant, insulin-dependent diabetic patients, initial good control can usually be obtained on CSII after 3 - 4 days of dosage adjustment on a metabolic ward,25'27 although full adjustment and stabilization occasionally take longer. Data on six diabetic patients of widely different duration of diabetes are shown in Table 2; it was possible to achieve near-normoglycemia in all six. The blood glucose control in one patient treated by CSII for 4 days (age 37 yr; duration of diabetes 7 yr; C-peptide, negative) is shown in Figure 2. During the first day he received 100% of his conventional therapy, but, because of symptomatic hypoglycemia on the first night, he was subsequently changed to an infusion rate of 79% of usual treat-

Patient no.

Age (yr)

1 3

34 39 23

4

23

5

16 20

2

6

24 22 0.1

0.2 10 15

Conventional treatment 20 U SOL *.8USOL/

15

I

10 5 0

16

20

24

8

12

16

Day 1 s.c. infusion hi 2-1 u/h 100%

E20 6

-1-6u/h79%

or 15 in

o ^ 10 o 5

3 Day 3 s.c. infusion l-6u/h

B

8

|L

i

12

16

E |

20

U 8 12 Time, h

16

20

2U

U

8

FIG. 2. Plasma glucose control in a patient during conventional treatment (upper panel) and during 4 days of CSII (lower 2 panels). CSII was initiated at a basal rate of 2.1 U/h, total of 100% of usual therapy but reduced to 1.6 U/h, 79% of usual therapy on day 2. B = breakfast, L = lunch, E = evening meal. (From: Diabetologia 16: 385-89,

1979.)

ment. On this regime, near-normoglycemia (most values between 4 and 10 mmol/L) was obtained and maintained. Other groups report similarly good short-term control by CSII. Tamborlane et al.22'35 obtained final mean daily blood glucose values from 4-2 to 5.2 mmol/L in six diabetic patients infused for 14 days. Experience of long-term CSII under outpatient conditions is now being reported.27'30'36'37 We have studied six patients treated for up to 4 mo.27 Two were newly diagnosed diabetic patients who had presented with classic symptoms and heavy ketonuria. Table 3 shows that good control—as assessed by Dextrostix readings, which were usually measured four times TABLE 3

TABLE 2 Plasma glucose control during hospital-stabilized CSII Duration of diabetes (yr)

20r

Duration of infusion and blood glucose control during long-term CSII at home (blood glucose was estimated by Dextrostix and a reflectance meter)

Mean ± SD plasma glucose after initial stabilization on CSII (mmol/L) 4.6 dt 7.8 dt 4.7 dt 6.8 dt 3.9 dt 5.1 dt

1.6 2.7 1.6 2.8 1.2 2.6

Patient no. 1

2 3

4 5 6

Duration of infusion (days)

53 57 48 51 111 80

DIABETES CARE, VOL. 3 NO. 2, MARCH-APRIL 1980

Mean ± SD blood glucose (mmol/L) 6.6 7.5 4.8 6.2

± ± ± ±

1.8 1.6 1.6 2.2

5.9 ± 2.0 6.1 ± 2.1

Values at 13.9 mmol/L or greater (%)

5.7 5.4 0 3.6 1.6 2.5

293

CONTINUOUS SUBCUTANEOUS INSULIN INFUSION/J. C. PICKUP AND ASSOCIATES

TABLE 4 Daily mean plasma glucose levels and M-values on conventional therapy and CSII Mean plasma glucose (mmol/L) no.

Conventional

CSII

Conventional

CSII

1

14.3 12.0 10.9

8.2 6.0 8.4 6.5 5.5

142 81

32 28 36 8.4 6.5

2 3 Duration of infusion (days)

FIG. 3. Mean daily blood glucose values (Dextrostix) during 111 days of home CSIJ (lower panel); dotted lines, 5 and 10 mrnollL levels. Upper panel, daily insulin infusion dose shown as a percentage of usual conventional insulin dosage. Dotted line: 100%. daily—was possible under home conditions. Glucose oxidase reagent strips, and many of the reflectance meters used to read them, rapidly diminish in accuracy at about 250 mg/dl (13.9 mmol/L) or more, so it was encouraging that, at worst, only 5.7% of readings were at this upper limit and, in some cases, none at all. The blood glucose control (individual Dextrostix values) in a patient treated for 111 days by CSII is shown in Figure 3. Tamborlane et al.36 and Champion et al.30 have also treated patients at home for many months; good control has again been sustained. Glycosylated hemoglobin is reduced by a period of strict CSII-induced control. After 14 days, CSII values were reduced from 15.3 ± 2.1% to 12.4 ± 1.5%22 and, in two patients treated for several months,26 from 14.9 and 13.9% to 6.5 and 11.4% (normal range 6-10.5%).

Mean M- value

4 5

61 13.2

79 85 109

Patients were fully ambulant but treated in hospital.

during a 15-h fast, even in response to exercise on a bicycle ergometer. The Yale group39 also have exercised patients during CSII, but at 90 min after lunch and 2 h after the preprandial insulin boost. Plasma glucose before and after exercise was not significantly different from that in nondiabetic subjects. Most investigators employ abdominally implanted cannulae. Infusion into the subcutaneous tissue of the thigh, however, does not apparently cause exercise hypoglycemia (A. Schiffrin, personal communication), presumably because the rate of access to the circulation is limited by the infuser rather than by factors that mobilize a subcutaneous depot of hormone. INTERMEDIARY METABOLITE CONTROL

lthough diabetes mellitus is historically and clinically a disease of disordered glucose metabolism, the concentration of many other intermediary metabolites is equally disordered.40 Consequently, any definition of metabolic control will encompass a variety HYPOGLYCEMIA of metabolities, any or all of which may play a pathogenic It has been a consistent finding that symptomatic hypoglyce- role in diabetic complications. It is a plausible contention mia is not a problem during CSII25'27 as it is not with intrave- that normalization of glycemia will be attended by regression nous infusion systems. In fact, it seems less frequent than of other linked metabolites to normal, though, clearly, this with conventional depot therapy, particularly when stren- supposition requires testing. In patients treated for 24 h by uous efforts are made to obviate hyperglycemia. The M- CSII, blood levels of lactate, pyruvate, 3-hydroxybutyrate, 26 value,2'38 a derivation of blood glucose deviations above or and alanine were also brought closer to nondiabetic values. below an arbitrary standard (say 80 mg/dl) with extra weight- As an example, blood lactate levels in seven diabetic patreated by CSII are shown in Figure 4- Tamborlane et ing for hypoglycemia, may be calculated for CSII patients and tients 22 indicates the lessened oscillation of glycemia.24'25 Table 4 al. have shown reduction toward normal of fasting values of shows the mean daily M-value and plasma glucose in five pa- plasma cholesterol, triglyceride, free fatty acids, and fasting tients on their hospital-stabilized CSII regime. The range of and postprandial branched chain amino acids by a few days M-values in nondiabetics is 0.1-12. The values were re- of CSII. Total normality cannot be claimed, at least in part duced in every case during CSII compared with conventional because its definition is not clear. However, the question is not whether it is complete but whether it is enough. depot therapy in the same subjects. The effect of subcutaneous infusion and strict control on Exercise does not cause significant hypoglycemia during CSII. One of our long-term patients was a keen sportsman the associated, probably secondary, hormonal abnormalities undertaking horseriding, squash, golf, etc., during infusion. of diabetes (e.g., growth hormone, glucagon, catecholThe Canadian group30 studied nine diabetic patients who re- amines) has yet to be fully investigated; however, it has been mained fasting during the basal CSII delivery and then un- shown that the increased plasma levels of growth hormone, dertook vigorous exercise. Mean plasma glucose did not fall epinephrine, and norepinephrine in response to exercise in

294

DIABETES CARE, VOL. 3 NO. 2, MARCH-APRIL 1980

CONTINUOUS SUBCUTANEOUS INSULIN INFUSION/J. C. PICKUP AND ASSOCIATES

•^

L

of high specific activity confirm the approximate timing of meal-time free insulin peaks.43

1-

CONTROL AND ALBUMINURIA

About 50% of nonketotic, conventionally treated insulindependent diabetic patients without clinical proteinuria (i.e., Albustix-negative) have a raised excretion of urinary § oJ Midnight 4 A M 8AM Noon 4PM 8PM albumin as detected by sensitive radioimmunoassay.44 The proportion with microalbuminuria apparently drops in paClock time tients with known duration of diabetes greater than 20 yr or FIG. 4- Blood lactate concentrations (mean ± SD) during 24 h ofCSH so, possibly because of the premature mortality and exclusion in seven insulin-dependent diabetic patients ( • — • ) . Dotted line: mean of patients with clinical proteinuria. The natural history of blood lactate for nondiabetic subjects. Shaded area: mean blood lactate ± 2 SD for nondiabetic subjects. (From: Lancet 1: 1255-58, 1979.) such moderate increases in albumin excretion is unknown— i.e., whether it predicts progression to clinical nephropathy remains to be established. Recent evidence, however, would the diabetic patient are significantly reduced and are compa- suggest that increased leakage of macromolecules through rable to those in nondiabetic subjects after 14 days of CSII.39 the glomerular capillary wall and their accumulation in the Failure to return glucose metabolities and hormones to nor- glomerular basement membrane and mesangial regions might development of nodular or diffuse glomal after short-term treatment with closed-loop techniques contribute to the45 later 47 such as the AEP cannot be taken to predict the longer-term merulosclerosis. " We recently treated, by CSII, seven long-standing diaeffectiveness of CSII. betic patients who were deliberately chosen as having elevated microalbuminuria during conventional treatment.44 PHARMACOKINETICS After only 1-3 days of strict control with CSII (plasma gluPlasma free insulin profiles during CSII42 measured by the cose 4.2 ± 1.5 to 6.5 ± 3.3 mmol/L, mean ± SD), the uripolyethylene glycol precipitation method indicate that the nary albumin was reduced in all and normalized in three. basal and fasting levels are about 10-15 mU/L (Figure 5), Simultaneous assay of urinary /32 microglobulin (an index of not far removed from nondiabetic concentrations and much tubular dysfunction) suggests that the change in urinary allower than those seen during short-term studies with the arti- bumin excretion resulted from reduction of glomerular capilficial endocrine pancreas.41 Peaks of 30-50 mU/L occur lary premeability and not from a reduction in tubular reababout 75 min42'57 after switching to the high rate. Recent sorption capacity for proteins. These patients who had work shows that the absorption pattern of a bolus of insulin suffered from diabetes from 6 to 33 yr presumably have gloinjected subcutaneously differs little from the same dose and merular basement membrane thickening, which is highly unvolume infused subcutaneously over 17 min.42 CSII creates likely to have changed structurally in the short time of the in the diabetic patient a "best fit" of a meal-time circulating study. Functional factors, which are potentially reversible by insulin profile and the pattern of glycemia and insulinemia metabolic near-normalization, must still, therefore, govern after a meal in nondiabetic subjects. Preliminary studies in part capillary permeability in patients with diabetes of using continuous subcutaneous infusions of tritiated insulin long duration and with probable structural change. CONTROL AND DIABETIC RETINOPATHY

Two patients whom we have recently treated by long-term CSII have had severe proliferative or preproliferative retig 50 nopathy.48 One was a 20-yr-old with a 15-yr history of insulintreated diabetes, previously poorly controlled on a single mixed dose of soluble and protamine zinc insulins. She •* 30 presented to the Hammersmith Hospital, London, as a pos20 sible candidate for hypophysectomy. Fluorescein angiograms (Figure 6) showed the typical picture of preproliferative reti10 nopathy: marked capillary dilatation and leakage and areas 0 of capillary nonperfusion and capillary looping, indicative of 08 12 16 20 24 04 08 early proliferative vessels. After 3 mo of CSII-induced, Clock time.hr control (blood glucose, mean ± SD, 6.1 ± 2.1 of FIG. 5. Plasma free insulin levels in seven diabetic patients during 24 himproved CSII. Error bars represent SD. Breakfast was given at 0830, lunch atmmol/L), there were improvement in vision and remarkable 1230, and evening meal at 1830 h. Pump was turned to the high rate 30 reduction of retinovascular premeability to dye, revascularization of previously nonperfused areas (by normal vessels), and min before each of these meab.

DIABETES CARE, VOL. 3 NO. 2, MARCH-APRIL 1980

295

CONTINUOUS SUBCUTANEOUS INSULIN INFUSION/J. C. PICKUP AND ASSOCIATES

FIG. 6. Early phase fluorescein angiograms of the left eye, below the macula, in a diabetic patient during conventional treatment (left) and after 3 mo CSU (right). less marked looping. Such amelioration had never been observed so quickly except after hypophysectomy. Similar regression of retinopathy following prolonged intravenous, open-loop near-normalization has recently been reported by Irsigler et al.49 Progress of proliferative retinopathy in one patient was also arrested during the period of CSII, although intercurrent photocoagulation was performed. These cases do not indicate that other ways of achieving good control would not have been equally successful or that other cases of severe retinopathy will respond equally well; however, they do suggest that, even in its late stages, diabetic retinopathy may still be responsive to the imposition of dramatically improved degrees of metabolic control. The power of good control may thus not be confined to primary prevention or the reversal of only the earliest manifestation of diabetic microangiopathy; more patients with threatened vision or advancing renal disease must be studied.

NEWLY DIAGNOSED DIABETES

I

t has been suggested that a period of normoglycemia at the diagnosis of insulin-dependent diabetes may rescue some B-cell function by breaking the vicious circle of the hyperglycemia of insulin secretory inadequacy and the B-cell overstimulation. Mirouze et al. 50 were able to induce an increased frequency and duration of "honeymoon" remissions in newly diagnosed diabetic patients by a few days of initial control on an artificial pancreas. CSII should offer longer periods of good initial control. Two of our long-term outpatient cases were newly diagnosed ketonuric diabetic patients, and both enjoyed progressive reduction in insulin dose during the infusion while near-normoglycemia was

296

maintained.27 One of these patients was completely withdrawn from insulin therapy after 48 days of CSII and remained well controlled for a further 6 wk on diet treatment alone. While at present it seems unlikely that any more than temporary "remissions" of diabetes can be achieved in this way, it seems possible that some long-term islet secretory capacity may be conserved and that even small degrees of secretory capacity may improve control51 and therefore be helpful in reducing or delaying the eventual development of complications. PREGNANCY AND CONCEPTION

CSII has also been used during pregnancy in insulin-dependent diabetic women.23 Although greatly improved control can often be achieved in the commonly highly motivated pregnant patient by the more conscientious application of conventional treatment (and perhaps by the suspected amelioration of the diabetic state in pregnancy), some women do experience difficulty in obtaining adequate improvement, and CSII can probably be of assistance here. An as yet theoretical application of CSII is in the period of preparation for conception. There are strong suggestions that the residual problem of congenital abnormality in the offspring of diabetic mothers may be related to inadequate metabolic control during the first trimester of pregnancy, critical for organogenesis. Optimum control should perhaps be achieved before conception and maintained throughout pregnancy, and CSII could contribute to this. CHANGES IN INSULIN DOSAGE

In patients meeting prevailing criteria of good control on conventional therapy it is commonly found that, during

DIABETES CARE, VOL. 3 NO. 2, MARCH-APRIL 1980

CONTINUOUS SUBCUTANEOUS INSULIN INFUSION/J. C. PICKUP AND ASSOCIATES

CSII, better blood glucose control is achieved within a few days on about 80% of the preceding total daily dose.25*52 This may be explained by better absorption of the small volumes infused; large depots of subcutaneous insulin may be more subject to degradative enzymes and erratic absorption. Also, the more appropriate distribution of meal-time peaks and basal delivery may be physiologically more efficient, avoiding, for example, the Somogyi effect of hypoglycemia and overshoot hypoglycemia. The insulin dose changes in six patients during long-term

200 Casei 100

CSII are shown in Figure 7. Only one diabetic patient (no. 6) required consistently more insulin on CSII than conventional therapy, probably reflecting prior undertreatment. Patient 5 is interesting in that the dosage gradually fell from the stabilized level on CSII of 92 U/day to about 35 U/day. This long-standing diabetic patient was previously treated by PZI and soluble insulin. Asplin et al.53 have shown that, when patients on high doses of traditional insulins are transferred to highly purified preparations, a reduction in dose often occurs, due perhaps to decrease in circulating anti-insulin antibodies. However, reduction of insulin per se may improve levels of control in conventionally treated diabetics, and reduction in insulin dose and improved control is a well known, nonspecific response to increased "caring." Dramatic dosage reduction in two newly diagnosed patients (see section above) is also shown (patients 3 and 4).

0 200

Case 2

BRITTLE DIABETES

W

e define "brittle" diabetes as a syndrome of wide, unpredictable swings in blood glucose (and often associated metabolites and hor100 mones) so severe as to seriously disrupt or even threaten a patient's life. This wild instability exists a even in the presence of constancy of treatment, in spite of strenuous efforts by the physician to gain improved control, and is seemingly unrelated to food, exercise, or timing of ina 100 sulin injection.54 As defined, the term describes a rare subset o Case 3 '•& of diabetic patients and should not be applied to patients O» n mismanaged or the more metabolically labile ordinary insulindependent diabetic patients (from whom, we consider, some or all brittle diabetic patients may differ only in degree). We became aware soon after the start of our CSII program that truly brittle patients are little improved by our system.24 For example, we described a 33-yr-old female of 23 yrs' duration of diabetes who had a range of daily blood glucose levels from 2.2 to 28.0 mmol/L on three injections of soluble insulin daily. After several days of attempted stabilization on CSII the range was 3.4-25.2 mmol/L.25 The three brittle patients studied by Kitabchi et al.55 over a longer period of time were also uncontrolled on CSII. Since intravenous insulin infusion usually succeeds in establishing control, we speculated that some brittle diabetic patients may exhibit irregular, unpredictable breakdown of insulin in the subcutaneous tissue or highly irregular access to the circulation from it, and that this might explain the failure to improve control by modifying the manner of subcutaneous delivery from depot to infusion. We have now treated several brittle diabetic patients by continuous intramuscular insulin infusion (i.e., bypassing the subcutaneous 0 10 20 30 40 50 60 70' 80 90 100 110 barrier), one of them for 9 mo (to date) under home conditions. Total daily insulin dose has dropped to about one-half Duration of infusion (days) of that used in prior subcutaneous therapy and, furthermore, blood glucose values become near-physiologic throughout FIG. 7. Insulin dosage changes during CSII, calculated as the percentage the day. The long-term suitability of intramuscular infusion, of total daily dose on prevalent conventional therapy or initial CSII dose. however, has yet to be evaluated. (From Lancet 2: 870-73, 1979.)

* o

DIABETES CARE, VOL. 3 NO. 2, MARCH-APRIL 1980

297

CONTINUOUS SUBCUTANEOUS INSULIN INFUSION/]. C. PICKUP AND ASSOCIATES

PROBLEMS WITH CSII

D

uring some episodes of intercurrent illness and in one patient during severe emotional stress (patients treated at home), blood glucose control was lost on CSII, and insulin requirements increased.27 Patients must be instructed, in precise terms, that in such circumstances extra insulin is required to regain control either by increasing concentration of insulin in the pump syringe or by injecting insulin at a separate site. Occasionally the subcutaneous cannula is accidentally withdrawn from the implantation site, and patients should be provided with an emergency set (e.g., "Butterfly" needle) and adequately trained for self-insert ion. Potentially the most dangerous time is during the night when cannula withdrawal could result in many hours without insulin. Of course, these patients have little or no exogenous or endogenous insulin reserve, and blood glucose can rise quite rapidly. Champion et al.30 studied four patients after deliberately discontinuing CSII. The mean plasma glucose after 12 h was only 215 mg/dl (range 160—420), but more information will have to be obtained in a larger group of patients of defined C-peptide status, since the speed of development of ketoacidosis after insulin withdrawal is dependent on remaining Bcell function.56 The precipitating circumstance of infective, traumatic, or emotional stress may also contribute to faster rates of metabolic disintegration than seen in simple insulin withdrawal. Pump malfunction occurs from time to time and an audible warning of such breakdown should be incorporated in future models. A reduction in size and weight of infusers is essential to make them more acceptable to patients during prolonged use. It is the patient who, finally, requires to be metabolically monitored, and during the day this is readily achievable by self-monitoring of glycemia or of glycosuria. However, as indicated above, patient monitoring at night presents special problems, and approaches to this are currently being undertaken. At the present state of development, CSII requires clear understanding by the patient and round-the-clock availability of the supervising team. Several of the untoward events experienced have been attributable to failure of comprehension or action by the patients; in our view, all of them are rectifiable. The insulin-dependent diabetic patient is always "at risk" of metabolic disaster. We feel that, as used by us at present, this risk is not increased and, indeed, for a number of reasons, may in fact be reduced. CONCLUSIONS

The question of whether CSII is more effective than optimized conventional therapy is often raised. At one level, we would have accomplished our aim if it were possible to show the link between microangiopathy and strict control, no matter how such control was achieved. An answer to the control-complications debate will tell us not only about pathogenesis but also indicate the intensity with which we

298

must henceforth strive for good control with any system of insulin therapy. Nevertheless, there are some indications that CSII is superior to conventional injections at home, whilst being broadly similar in the hospital.30 Patient compliance is often better for CSII than for multiple injections, and this is likely to improve as pumps are further miniaturized and simplified. Two further advantages of CSII are the low frequency of symptomatic hypoglycemia and the flexibility of eating that it allows. With continued basal supply of insulin, meal times can be varied and complete fasting without hypoglycemia is theoretically possible. We are in a greatly improved position to vary insulin dose to fit the timing, composition, and magnitude of meals, and less compelled to fit the food pattern to the insulin dose schedule. Despite all this, CSII must remain a research tool for the near future. Patients recruited for its trial must be fully informed and cognizant, rigorously educated, demonstrably responsive to instruction, and strictly supervised by doctors with ready recourse to reparative hospital facilities. The development of smaller and more reliable devices and, above all, of regulatory metabolic sensor systems will allow the wider expansion of CSII with acceptable fail-safe standards to more general application in patient care. ACKNOWLEDGMENTS: We would like to express our appreciation and indebtedness to the British Insulin Manufacturers, the Medical Research Council, and the Minet Trust for financial support. Skillful technical assistance was provided by K. Kilbourn, B. Lloyd, S. Pickup, E. Regina, M. Stimmler, P. Smythe, and D. Williams. R. Blows constructed the special dual-rate infusers. We thank the sisters, staff nurses, and dietitians of the metabolic wards at Guy's Hospital and Hammersmith Hospital for their help and L. Smith and J. Gray for expert secretarial assistance. Finally, without the kindness and unfailing cooperation of our patients this study would have been impossible. From the Unit for Metabolic Medicine, Guy's Hospital Medical School, London; Endocrine Unit, Hammersmith Hospital, London; and National Institute for Medical Research, London, and Royal Victoria Infirmary, Newcastle-upon-Tyne, England. Address reprint requests to J. C. Pickup, Unit for Metabolic Medicine, Guy's Hospital Medical School, London, SE19RT, England. REFERENCES 1 Molnar, G. D., Taylor, W. F., and Ho, M. M.: Day-to-day variation of continuously monitored glycaemia: a further measure of diabetic instability. Diabetologia 8: 342-48, 1972. 2 Service, F. J., Molnar, G. D., Rosevear, J. W., Ackerman, E., Gatewood, L. C., and Taylor, W. F.: Mean amplitude of glycaemic excursions, a measure of diabetic instability. Diabetes 19: 644-55, 1970. 3 Engerman, R., Bloodworth, J. M. B., and Nelson, S.: Relationship of microvascular disease to metabolic control. Diabetes 26: 760-69, 1977.

DIABETES CARE, VOL. 3 NO. 2, MARCH-APRIL 1980

CONTINUOUS SUBCUTANEOUS INSULIN INFUSION/J. C. PICKUP AND ASSOCIATES

4 Tchobroutsky, G.: Relation of diabetic control to development M.: Continuous subcutaneous insulin infusion: good blood glucose of microvascular complications. Diabetologia 15: 143-52, 1978. control for up to 4 days. Diabetologia 16: 3 8 5 - 8 9 , 1979. 5 26 Matas, A. J., Sutherland, D. E., and Najarian, J. S.: Current Pickup, J. C , Keen, H. ( Parsons, J. A . , Alberti, K. G. M. M., status of islet and pancreas transplantation in diabetes. Diabetes 25: and Rowe, A. S.: Continuous subcutaneous insulin infusion: im785-95, 1976. proved blood glucose and intermediary metabolite control in diabet6 Sonksen, P. H., Judd, S. L., and Lowy, C : Home monitoring of ics. Lancet 1: 1255-58, 1979. 27 blood glucose. Lancet J: 729-32, 1978. Pickup, J. C , White, M. C , Keen, H., Kohner, E. M., Par7 Walford, S., Gale, E. A. M., Allison, S. P., and Tattersall, R. sons, J. A . , and Alberti, K. G. M. M.: Long-term continuous subB.: Self-monitoring of blood-glucose. Lancet 1: 732-35, 1978. cutaneous insulin infusion in diabetics at home. Lancet 2: 8 7 0 - 7 3 , 8 Santiago, J. L., Clemens, A. H., Clarke, W. L., andKipnis, D. 1979. 28 M.: Closed-loop and open-loop devices for blood glucose control in Binder, C : Absorption of injected insulin. A clinical pharmanormal and diabetic subjects. Diabetes 28: 71-81, 1979. cological study. Acta Pharmacol. Toxicol. 27 (Suppl. 2): 1-87, 9 Metcalf, J.: The administration of insulin by continuous injec- 1967. 29 Hepp, K. D., Renner, R., Franetzki, M., and Mehnert, H.: tion. M. B. Thesis, University of Cambridge, 1934. 10 Pickup, J. C , Keen, H., Parsons, J. A., and Alberti, K. G. M. Substitution of islet-cell function with "open-loop" insulin infusion M.: The use of continuous subcutaneous insulin infusion to achieve systems. In Advances in Experimental Medicine and Biology, Vol. normoglycaemia in diabetic patients. Diabetologia 13: 425A, 1977. 119. Camerini-Davalos, R. A., and Hanover, B., Eds. New York, 11 Pickup, J. O , and Keen, H.: Continuous subcutaneous insulin Plenum Press, 1979, pp. 485-489. 30 Champion, M. C , Shepherd, G. A. A . , Rodger, N . W . , and infusion: a developing tool in diabetes research. Diabetologia 18: Dupre, J.: Continuous subcutaneous infusion of insulin in the man1-4, 1980. 12 Albisser, A. M., Leibel, B. S., Ewart, T. G., Davidovac, Z., agement of diabetes mellitus. Diabetes 29: 2 0 6 - 1 2 , 1980. 31 Job, D., Eschwege, E., Guyot-Argenton, C , Aubry, J. A . , Botz, C. K., Zingg, W., Schipper, H., and Gander, R.: Clinical control of diabetes by the artificial pancreas. Diabetes 23: 396-404, and Tchobroutsky, G.: Effect of multiple daily injections on the course of diabetic retinopathy. Diabetes 25: 4 6 3 - 6 9 , 1976. 1974. 32 13 Jarrett, R. J., and Keen, H.: Hyperglycaemia and diabetes melAlbisser, A. M., and Leibel, B. S.: The artificial pancreas. litus. Lancet 2: 1009-12, 1976. Clin. Endocrinol. Metab. 6: 437-47, 1977. 14 33 Slama, G., Hautcouverture, M., Assan, R., and TchoDorf, A . , Ballintine, E. J., Bennett, P. H., and Miller, M.: broutsky, G.: One to five days of continuous insulin infusion on Retinopathy in Pima Indians. Diabetes 25: 5 5 4 - 6 0 , 1976. 34 seven diabetic patients. Diabetes 23: 732-38, 1974Rushforth, N . B., Miller, M., and Bennett, P. H.: Fasting and 15 Deckert, T., and Ubrup, B.: Regulation of brittle diabetes by a two-hour post-load glucose levels for the diagnosis of diabetes. Diapreplanned infusion programme. Diabetologia 12: 573-79, 1976. betologia 16: 3 7 3 - 7 9 , 1979. 16 35 Hepp, K. D., Renner, R., Funcke, H. J., Mehnert, H., HaerTamborlane, W. V., Sherwin, R. S., Genel, M., and Felig, ten, R., and Kresse, H.: Glucose homeostasis under continuous in- P.: Reduction to normal of plasma glucose in juvenile diabetes by travenous insulin therapy in diabetics. Horm. Metab. Res. (Suppl.) subcutaneous administration of insulin with a protable infusion 7: 72-76, 1977. pump. N . Engl. J. Med. 300: 5 7 3 - 7 8 , 1979. 17 36 Genuth, S., and Martin, P.: Control of hyperglycemia in adult Tamborlane, W . V., personal communication. 37 diabetics by pulsed insulin delivery. Diabetes 26: 571-81, 1977. Riley, W . J., Silverstein, J. H., and Rosenbloom, A. L.: Am18 Service, F. J.: Normalization of plasma glucose of unstable dia- bulatory diabetes management by a pulse of subcutaneous insulin betes: studies under ambulatory, fed conditions with pumped intra- delivered by a portable pump: preliminary report. Diabetes Care 2: venous insulin. J. Lab. Clin. Med. 91: 480-89, 1978. 2 7 2 - 7 4 , 1979. 19 38 Hanna, A. K., Minuk, H. L , Zinman, B., Marliss, E. B., EllSchlichtkrull, J., Munck, O . , and Jersild, M.: T h e M-value, man, J., Leibel, B. S., and Albisser, A. M.: A portable system for an index of blood sugar control in diabetics. Acta Med. Scand. 177: continuous intravenous insulin delivery: characteristics and results 9 5 - 1 0 2 , 1965. 39 in human diabetics. Clin. Res. 26: 630A, 1978. Tamborlane, W. V., Sherwin, R. S., Koivisto, V., Hendler, 20 Irsigler, K., and Kritz, H.: Long-term continuous intravenous R., Genel, M., and Felig, P.: Normalization of the growth hormone insulin therapy with a portable insulin dosage-regulating apparatus. and catecholamine response to exercise in juvenile-onset diabetic Diabetes 28: 1 9 6 - 2 0 3 , 1979. subjects treated with a portable insulin infusion pump. Diabetes 28: 21 Parsons, J. A . , Rothwell, D., and Sharpe, J. E.: A miniature 785-88, 1979. 40 syringe pump for continuous administration of drugs and hormones: A l b e r t i , K. G . M . M . , D o r n h o r s t , A . , a n d R o w e , A . S.: M e t a the Mill Hill Infuser. Lancet 1: 7 7 - 7 8 , 1977. bolic r h y t h m s in n o r m a l a n d diabetic m a n . Studies in insulin22 Tamborlane, W. V., Sherwin, R. S", Genel, M., and Felig, P.: treated diabetes. Isr. J. M e d . Sci. 11: 5 7 1 - 8 0 , 1975. 41 Restoration of normal lipid and amino acid metabolism in diabetic Horowitz, D. L., G o n e n , B . , Zeidler, A . , Langer, B . , a n d patients with a portable insulin-infusion pump. Lancet 1: 1258—61, R o d m a n , D . : A n "artificial b e t a cell' for c o n t r o l of diabetes: c o m 1979. parison of glucose a n d free insulin levels w i t h those a c h i e v e d by sub23 Potter, J. M . , Reckless, J. P. D . , a n d Cullen, D . R.: T h e c o n - c u t a n e o u s insulin. Diabetes 2 6 (Suppl. 1): 3 7 6 , 1977. 42 trol of glucose a n d intermediary metabolites in pregnant diabetic paH o m e , P . , Pickup, J. C , a n d K e e n H . , unpublished d a t a . 43 tients on different insulin regimens. Clin. Sci. Mol. Med. 55: 4, Pickup, J. D . , H a l b a n , P . , Philippe, J., Offord, R. E., a n d 1978. Keen, H . , unpublished data. 24 44 Pickup, J. C , Keen, H. ( Parsons, J. A . , and Alberti, K. G. M. Viberti, G . C , Pickup, J. C , Jarrett, R. J., a n d K e e n , H . : EfM.: Continuous subcutaneous insulin infusion: an approach to fect of c o n t r o l of blood glucose o n urinary e x c r e t i o n of a l b u m i n a n d achieving normoglycaemia. Br. Med. J. 1: 2 0 4 - 0 7 , 1978. /3 2 microglobulin in i n s u l i n - d e p e n d e n t d i a b e t e s . N . Engl. J. M e d . 25 Pickup, J. C , Keen, H., Parsons, J. A . , and Alberti, K. G. M. 300: 638-41, 1979.

DIABETES CARE, VOL. 3 NO. 2, MARCH-APRIL 1980

299

CONTINUOUS SUBCUTANEOUS INSULIN INFUSION/J. C. PICKUP AND ASSOCIATES

45

Mauer, S. M . , Fish, A . J., Blau, E. B . , and Michael, A . F.: T h e glomerular mesangium. I. Kinetic studies of macromolecular uptake in normal and nephrotic rats. J. Clin. Invest. 51: 1092— 1101, 1972. 48 Hoyer, J. R., Elema, J. D., and Vernier, R. L : Unilateral renal disease in the rat. II. Glomerular mesangial uptake of colloid carbon in unilateral aminonucleoside nephrosis and nephrotoxic serum nephritis. Lab. Invest. 34: 2 5 0 - 5 5 , 1976. 47 Velosa, J. A . , Glasser, R. J., Nevins, T . E., and Michael, A . F.: Experimental model of local sclerosis. II. Correlation with immunopathological changes, macromolecular kinetics and polyanion loss. Lab. Invest. 36: 5 2 7 - 3 4 , 1977. 48 W h i t e , M . C , Kohner, E. M . , Pickup, J. C , and Keen, H . : Improvement in retinal vascular changes o n continuous subcutaneous insulin infusion (CSII). Presented at British Diabetic Association Meeting, September 1979. 49 Irsigler, K., Kritz, H . , Najemnik, C , and Freyler, H.: Reversal of florid diabetic retinopathy. Lancet 2: 1068, 1979. 50 Mirouze, J., Selam, J. L., Pham, T . C , Mendoza, E., a n d O r setti, A . : Sustained insulin-induced remissions of juvenile diabetes by means of a n external artificial pancreas. Diabetologia 14: 223— 27, 1978. 51 G o n e n , B., Goldman, J., Baldwin, D . , Goldberg, R. B., Ryan, W . G . , Blix, P. M . , Schanzlin, D . , Fritz, K. J., and Rubin-

300

stein, A . H . : Metabolic control in diabetic patients. Effect of insulin secretory reserve (measured by plasma C-peptide levels) and circulating antibodies. Diabetes 28: 7 4 9 - 5 3 , 1979. 52 Pickup, J. C , Keen, H . , Stevenson, R. W . , Parsons, J. A . , Alberti, K. G . M . M . , W h i t e , M . , and Kohner, E. M . : Insulin via continuous subcutaneous infusion. Lancet 2: 9 8 8 - 8 9 , 1978. 53 Asplin, C . M . , Hartog, M . , and Goldie, D. J.: Change of insulin dosage, circulating free and bound insulin and insulin antibodies on transferring diabetics from conventional to highly purified porcine insulin. Diabetologia 14: 9 9 - 1 0 5 , 1978. 54 Tattersall, R. B.: Brittle diabetes. Clin. Endocrinol. Metab. 6:

403-19, 1977. 55

Kitabchi, A . E . , Fisher, J. N . , Burghen, G . A . , Gaylord, M . S., and Blank, N.: Evaluation of a portable insulin infusion pump for outpatient management of brittle diabetes. Diabetes Care 2: 421-24, 1979. 58 Madsbad, S., Alberti, K. G. M. M., Binder, C , Burrin, J. M., Faber, O. K., Krarup, T., and Regeur, L: Role of residual insulin secretion in protecting against ketoacidosis in insulin-dependent diabetes. Br. Med. J. 2: 1257-59, 1979. 57 Slama, G., Buu, K. N. P., Tchobroutsky, G., Delage, A., and Desplanque, N.: Plasma insulin and C-peptide levels during continuous subcutaneous insulin infusions. Diabetes Care 2: 251-55, 1979.

DIABETES CARE, VOL. 3 NO. 2, MARCH-APRIL 1980