LONG-TERM CONTINUOUS SUBCUTANEOUS INSULIN INFUSION IN DIABETICS AT HOME

870

LONG-TERM CONTINUOUS SUBCUTANEOUS INSULIN INFUSION IN DIABETICS AT HOME M. C. WHITE H. KEEN J. A. PARSONS K. G. M. M. ALBERTI

J. C. PICKUP

Unit for Metabolic Medicine, Guy’s Hospital Medical School, London; Endocrine Unit, Hammersmith Hospital, London; National Institute for Medical Research, London; and Department of Clinical Biochemistry, Royal Victoria Infirmary, Newcastle upon Tyne

6 insulin-requiring diabetics were treated at home with continuous, longterm, dual-rate subcutaneous insulin infusion (CSII) by means of a portable syringe pump. The duration of infusion was 48-111 days. Patients were initially stabilised on CSII in hospital and then allowed home, where capillary-blood glucose control was monitored by the patients with glucose-oxidase reagent strips. Patients diluted and changed their own insulin for the pump, adjusting the dose according to the control achieved. Mean (±SD) blood-glucose values ranged from 4·8±1·6 to 7·5±1·6 mmol/l. In 1 longstanding diabetic insulin requirements fell from 92 U/day to about 35 U/day on CSII. 2 newly diagnosed juvenile-onset diabetics were also infused: in 1 patient requirements fell to zero after 48 days and in the other the dose fell to 14 U/day after 51 days of CSII. No cannula-site infection or significant palpable lipodystrophy was experienced. Patients treated with "open-loop" systems have little or no insulin

Summary

potential loss of control—for example, during intercurrent illness—demands careful metabolic monitoring and prompt correction.

reserve:

Introduction THERE is increasing evidence for the importance, of good diabetic control in delaying the onset and reducing the severity of diabetic microangiopathy.’,’ Conclusive proof, however, is lacking because even the best control achieved by conventional methods cannot maintain near-normal metabolism for long periods. We have developed the technique of low-volume, continuous, subcutaneous insulin infusion (CSII) using a dual-rate portable syringe pump3,4 to achieve long-term strict diabetic control. Use of the subcutaneous route avoids the thrombosis and infection associated with long-term intravenous infusion. Like other "open-loop" devices, the system seeks the simplicity of good control without glycaemia-regulated feedback. We have shown that CSII established near-normoglycaemia in fully ambulant insulin-dependent diabetics treated for up to 4 daysS but supervised in hospital. Circadian profiles of the major intermediary metabolites, lactate, pyruvate, 3-hydroxybutyrate, and alanine, were also brought close to the non-diabetic state during CSII.6 Other workers7,8 have confirmed and extended these findings. We have now investigated outpatient use of CSII for up to 4 months to establish the feasibility of long-term use of such electromechanical. devices in diabetes research.

cal features are summarised in table I. 2 of the patients (nos. 5 and 6) were attending the Hammersmith Hospital retinopathy clinic with severe and progressive diabetic retinopathy. An account of their retinal response to CSII will be presented elsewhere. Patients were admitted to a metabolic ward. On day 1 blood-glucose control was monitored on the patient’s conventional injection regimen (or without therapy in the case of newly diagnosed diabetics, nos. 3 and 4) to assess current control and as a guide to the rate of infusion during subsequent CSII. Blood was withdrawn through an indwelling venous cannula (kept patent with 0-154 mol/1 saline injections) usually every hour during the day and at least every 2 h at night; plasma glucose was measured with the glucose-oxidase method by means of an Analox GM5 (Analox, Oxford) or Beckman automatic analyser. Continuous subcutaneous insulin infusion was started on the evening of day 2. The patients were initially infused with the same total daily dose of insulin (including meal-time increases) as on conventional treatment. ’Actrapid’ monocomponent insulin (Novo Industri) was diluted with 0-154 mola saline, as necessary, for each patient. Insulin was delivered from the pump through a long-term implanted nylon cannula of internal diameter 0.2 mm (Portex Ltd., Hythe, Kent). The procedure for insertion of the cannula in the subcutaneous tissue of the abdomen has been described

previously.4 Pump System A single-rate

Mill Hill

infuser,9 model 1001A (Muirhead electronically adapted to deliver at two rates, 50 }jJ/h and 400 .1/h. The higher rate was activated by the patient (30 min before breakfast, lunch, and evening meal) by pressing a guarded button on the side of the

Ltd., Beckenham, Kent)

pump, the rate automatically returning to the basal level after 17 min. The total volume infused per day was about 13 ml, depending on the number of meal-time boosts (see below). The infuser is powered by a mercury battery and accepts 3 ml dis-

posable plastic syringes. The hub of the delivery cannula was secured to the syringe by a screw-fitting Luer lock. The infuser weighed 300 g, measured 144 mm x 72 mm x 23 mm, and was worn by the patient in a shoulder harness or attached to a

belt around the waist.

Infusion Strategy The

daily

infusion

rate was

insulin-requiring nature

diabetics volunteered for the study, the was carefully explained. Their clini-

of which

or

decreased

by

to the blood-glucose control achieved and until the range of values was between 5 and 10 mmol/1 throughout 24 h. TABLE I-CLINICAL FEATURES OF PATIENTS

Patients and Methods 6

increased

10-20% steps during the first few days in hospital, according

‘

research

was

NPH=Isophane PZI=Protamine zinc insulin

871 While in

hospital patients

were

taught how

to

operate the

infusion pump and given instructions for emergency action in event of system failures. They were shown how to replace and refill the infuser syringes daily with diluted insulin, mixing appropriate and specified quantities of 40 U/ml or 80 U/ml actrapid insulin with sterile 0-154 mol/1 saline to obtain the correct concentration for the dose rate. The battery was changed weekly, irrespective of charge-state. Patients were taught to estimate capillary-blood glucose concentrations with glucose-oxidase reagent strips (’Dextrostix’, Ames-Miles) by comparison with the colour chart provided or by measurement of the strip colour in a reflectance meter (’Eyetone’, Ames-Miles). Measurements were made when possible fasting, before lunch, before the evening meal, and before going to bed. If the blood glucose rose to 11-15 mmol/1 patients were instructed to give an extra 17 min highrate infusion; if blood glucose rose above 15 mmol/1 patients were instructed to inject actrapid insulin subcutaneously by conventional means and to telephone for advice at any time of

the

night or day. Patients

outpatients in hospital approximately fortnight for inspection of the cannula implantation site, review of blood-glucose profiles, supply of materials (insulin, saline, dextrostix), and measurement of glycosylated haemoglobin (Hb AJ. Percentage Hb Al was assayed by the micro ion-exchange column chromatographic method of Welch and Boucher,lO with hmmolysates from fresh blood prepared and run immediately or from washed packed cells after storage forldayat-20°C. Cannulae were re-implanted every 2-8 weeks, as necessary (see Results). The pump was disconnected from the cannula at bath-times, the cannula hub being covered with a cap or spiggot. Statistical analysis was by the Wilcoxon rank sum test for were seen as

once a

distribution-free data.

Fig. I-Blood-glucose profile during the last 3 weeks of CSII in patient S. Individual dextrostix readings were used to construct the curve. Dotted lines, 5 and 10 nuncio levels.

lowered in all patients while on CSII, so in 5 subjects. The variability of values about the mean was greatly reduced in the

Mean levels

remaining case (no. 2). Home dextrostix measurements are recorded in table III, with the number of readings per day and percentage of readings which reached or exceeded the upper limit on the chart (13-9 mmol/1, 250 mg/dl). The overall means of the daily mean blood-glucose values during the infusions (48-111 days) were within the normoglycaemic range (table iv). The range of values in all patients (mean of lowest and highest on each day) was within the limits set for the study (i.e., below 10-11

mmola, 180-200 mg/dl). Fig. 1 demonstrates the blood-glucose profile during the last 3 weeks of CSII in a patient with long-standing diabetes (no. 5). Individual dextrostix values were used to construct

Results

Blood-glucose Control Table n shows the plasma-glucose concentrations in the 6 patients during conventional treatment (as measured in hospital before CSII was started) and after stabilisation on CSII (immediately before discharge home). TABLE II-COMPARISON OF PLASMA-GLUCOSE CONTROL DURING PREVALENT

INJECTION

TREATMENT AND DURING

HOSPITAL-STABILISED CSII

were

statistically significantly

the

curve.

The concentration exceeded 10

mmol/1 only once (11 - 8 mmol/1) during this time and the patient experienced mild hypoglycaemic symptoms on one occasion (day 93). Fig. 2 shows the blood-glucose profile during the last 3 weeks of CSII in a patient with newly diagnosed diabetes (no. 4). Dispersion of values was slightly greater in this patient. Neither biochemical nor symptomatic hypoglycaemia occurred. Glycosylated haemoglobin was assayed before and after CSII in patients 3 and 4. Values on conventional therapy were 14.9% and 13.9% and at the termination of infusion were 6-5% and 11.4% (normal range 6.0-10.0%). Dose Changes

Fig.

3 shows the

CSII, calculated

in daily insulin dose during percentage of the dose on conven-

changes

as a

TABLE IV-DURATION OF CSII AND BLOOD-GLUCOSE CONTROL

ACHIEVED

NS=not significant. TABLE III-HOME-MONITORING OF BLOOD GLUCOSE

*Calculated as the mean of all daily mean values. mean of the lowest and highest value for all the

Calculated as the days.

872

Subcutaneous Cannulœ

left in situ for 2-8 weeks at a time. In there visible inflammatory reaction at the implantation site, and culture of the cannula tips proved sterile in every case. No visible or palpable tissue reaction occurred at the infusion site. Cannulm

were

no case was

Patient Response

Fig. 2-Blood-glucose profile during the last 3 weeks of CSII in patient 4. Individual dextrostix readings were used. Dotted lines, 5 and 10 mmol/1 levels. tional therapy or as a percentage of initial CSII dose in those patients not previously treated with insulin (nos. 3 and 4). Only patient 6 required consistently more insulin during CSII than when on conventional treatment. Patient 5 was stabilised and sent home on an infusion dose of 92 U/day (165% of her usual therapy), but this gradually fell to 34-39 U/day (60-70%) over the course of about 65 days. In the newly diagnosed diabetics insulin dosage fell in patient 3 to zero after 48 days and in patient 4 to 14 units (44% of initial dose) after 51days. The dosage in patients 1 and 6 remained relatively constant.

All patients coped well with the system, quickly learning the techniques of blood-glucose self-monitoring, dosage adjustment, and syringe replacement. Daily activities were not substantially impaired; patients were able to bath regularly (see Patients and Methods) and

undertake moderate exercise. Patient 1 was a keen sportsman and took part in golf, horse-riding, squash, and swimming. He largely avoided exercise-induced hypoglycsemia by taking exercise before the meal-time boost of insulin and by spigotting the cannula and removing the pump during the exercise. 2 patients went on holiday with the system in situ. Problems

Intercurrent illness impaired control on some occasions and patients were instructed to inject subcutaneous actrapid insulin at a separate site if blood-glucose values exceeded 15 mmol/l. For example, an episode of abdominal pain (unknown origin) in patient 5 was treated with conventional subcutaneous injection of an extra 12 U and subsequently 20 U of insulin. Accidental removal of the cannula from the subcutaneous site also occurred on a number of occasions at the beginning of the study. This problem was later largely solved by improvement in the technique of securing the cannula with several loops underneath the covering waterproof skin plaster. Nevertheless, blood-glucose values rose rapidly during episodes of severe stress or illness or after accidental cannula withdrawal, and a brief period of conventional treatment was then needed to re-institute near-normoglycxmia. Four such episodes were observed in the 6 patients studied.

Discussion This study demonstrates the efficacy of CSII in maintaining long-term near-normoglycxmia in outpatient insulin-requiring diabetics. Our aim was to develop a method of establishing good diabetic control (blood-glucose concentrations below 10-11 mmol/1) in such patients, since there is epidemiological evidencell,lz that significant retinal microvascular abnormalities do not occur in subjects with blood-glucose concentrations below about 200 mg/dl (11 mmol/1). A potential difficulty in such a study is in the reliability of ascertaining glycasmic control during infusion at home. Glycosylated haemoglobin values can be used to support the impression of strict control, as here, but it only reflects the degree of "excess glycsemia"

period of time. Some information on variability of test values in both directions is also required. Regular self-monitoring of capillary-blood glucose samples with glucose-oxidase reagent strips appears to be a useful method and this can be improved with reflectance-meter estimation of the value. Direct reading of the strips is unreliable at blood-glucose concentrations of about 13.9 mmol/l (250 mg/dl) and above. Several types of reflec-

over a

Fig.

3-Insulin

dosage changes during CSII,

calculated

as

the

percentage of total daily dose pn prevalent conventional therapy or initial CSII dose.

Dotted line shows starting dose.

873 also be inaccurate at these levels.13 For recorded the number of individual values reaching or exceeding 13-9 mmol/l. The percentage of such values in each patient ranged from nil to only

tance meters can

this reason,

we

CEREBRAL BLOOD-FLOW AND VISCOSITY IN RELATIVE POLYCYTHÆMIA

Factors which could make dose

changes

necessary

during CSII have been discussed elsewhere. 14 The reduced insulin requirements in the 2 newly diagnosed diabetics are noteworthy because of the suggestion that the incidence and duration of remission in early diabetes might be increased by strict initial control. Mirouze and his co-workers1S used an artificial pancreas-i.e., a closed-loop device-for a mean of 5 days to induce remissions. CSII can probably offer longer periods of near-normoglycxmia, and further studies areneeded to assess possible recovery of B-cell function during

infusion. The dramatic fall in insulin dose in patient 5 (from 92 to 34-39 U/day) may have been the result of altered circulating insulin antibodies during this time. Asplin et al.16 have found that patients on high doses of traditional insulins may require substantially less when transferred to highly purified insulins. All of the patients in the present study were infused with highly purified porcine insulin (actrapid monocomponent). Only patient 6 required consistently more insulin during CSII than on conventional treatment, probably reflecting undertreatment by the latter regimen. In this trial of prototype devices we observed episodes of rapidly increasing blood-glucose values associated with some instances of intercurrent illness, severe emotional stress, or accidental withdrawal of the cannula. This may have been because patients treated with CSII retain a very small subcutaneous reserve pool of insulin compared with conventional depot injection treatment. Further, patients who have been well controlled for long periods may respond more briskly with counter-regulatory factors when stressed or when insulin is acutely withdrawn: this hypothesis needs testing. The greatest vigilance is needed under such circumstances. Open-loop insulin-delivery systems clearly require further technical development. A reduction in size and weight of the infusion devices will make them more acceptable to patients, and microelectronics could allow the present simplicity of use to be combined with flexibility in adjustment to individual needs. Audible warning of pump malfunction or misuse and the development of a "metabolic sensor" should in the future allow the setting of fail-safe standards which must be the ultimate criterion of good design and exemplary patient care in a system to be used by subjects with little or no insulin reserve.

We thank our patients for their enthusiastic support in these studies ; the sisters and staff nurses of the metabolic wards of Guy’s Hospital and Hammersmith Hospital for their invaluable help; and Mrs S. Pickup for technical assistance. This work was supported by a grant

from the British Insulin Manufacturers.

Requests for reprints should be addressed to J. C. P., Unit for Metabolic Medicine, Guy’s Hospital Medical School, London SE 1 9RT. REFERENCES

1. Cahill GF, Etzweiler DD, Freinkel N. Blood glucose control and diabetes. Diabetes 1976; 25:237-40. 2. Tchobroutsky G. Relation of diabetic control to development of microvascu-

lar complications. Diabetologia 1978; 15:143-52. 3

Pickup JC, Keen H, Parsons JA, Alberti KGMM. The use of continuous subcutaneous insulin infusion to achieve normoglycæmia in diabetic patients. Diabetologia 1977; 13:425.

G. H. DU BOULAY T. C. PEARSON LINDSAY SYMON E. ZILKHA

P. R. D. HUMPHREY JOHN MARSHALL R. W. Ross RUSSELL G. WETHERLEY-MEIN

5.7%.

Institute of Neurology, National Hospital, Queen Square, London, and Department of Hœmatology, St. Thomas’ Hospital, London

Cerebral blood flow (CBF), red-cell mass, and plasma volume were measured in 39 patients with venous hæmatocrit values in the range 0·47-0·58. Patients with true polycythæmia were thus excluded. The 39 patients studied fell into two groups—those with a measured red-cell mass in the high-normal range and those with normal red-cell-mass values but reduced plasma-volume values (the relativepolycythæmia, low-plasma-volume group). The mean CBF was low in both the high-normal red-cell-mass (HNRCM) group (45·8 ml/100 g/min) and low-plasmavolume (LPV) group (48·8 ml/100 g/min). The hæmatocrit in the HNRCM group was 0·504 and in the LPV group 0·513. In a control group of subjects with a mean hæmatocrit of 0·421 CBF was 68·6 ml/100 g/min. Venesection was associated with a significant rise in CBF in both groups of patients—to 59·7 ml/100 g/min in the HNRCM group and 65·0 ml/100 g/min in the LPV group. Whole-blood viscosity fell significantly in both groups. Both groups demonstrated significant inverse relationships between CBF and hæmatocrit and between CBF and blood viscosity.

Summary

Introduction THE classification of even

patients

considerable elevation of

in whom equivocal or haematocrit sug-

venous

Pickup JC, Keen H, Parsons JA, Alberti KGMM. Continuous subcutaneous insulin infusion: an approach to achieving normoglycæmia. Br Med J 1978; i: 204-07. 5. Pickup JC, Keen H, Parsons JA, Alberti KGMM. Continuous subcutaneous insulin infusion: good blood glucose control for up to 4 days. Diabetologia 1979;16:385-89. 6. Pickup JC, Keen H, Parsons JA, Alberti KGMM, Rowe AS. Continuous subcutaneous insulin infusion: improved blood-glucose and intermediarymetabolite control in diabetics. Lancet 1979; i: 1255-58. 7. Tamborlane WV, Sherwin RS, Genel M, Felig P. Reduction to normal of plasma glucose in juvenilediabetes by subcutaneous administration of insulin with a portable infusion pump. N Engl J Med 1979; 300: 573-78. 8. Tamborlane WV, Sherwin RS, Genel M, Felig P. Restoration of normal lipid and aminoacid metabolism in diabetic patients with a portable insulin4.

infusion pump. Lancet 1979; i: 1258-61. 9. Parsons JA, Rothwell D, Sharpe JE. A miniature syringe pump for continuous administration of drugs and hormones: the Mill Hill Infuser. Lancet 1977; i: 77-78. 10. Welch SG, Boucher BJ. A rapid micro-scale method for the measurement of

hæmoglobin A1(a+b+c). Diabetologia 1978; 14:209-11. 11. Rushforth NB, Miller M, Bennett PH. Fasting and two-hour post-load glucose levels for the diagnosis of diabetes. Diabetologia 1979; 16: 365-75. 12. Jarrett RJ, Keen H. Hyperglycæmia and diabetes mellitus. Lancet 1976; ii: 1009-12. 13. Borthwick LJ, Ross IS. Performance of blood glucose meters. Lancet 1979; i: 924. 14. Pickup JC, Keen H, Stevensen RW, et al. Insulin via continuous subcutaneous infusion. Lancet 1978; ii: 988-89. 15. Mirouze J, Selam JL, Pham TC, et al. Sustained insulin-induced remissions of juvenile diabetes by means of an external artificial pancreas. Diabetologia 1978; 14: 223-27. 16. Asplin CM, Hartog M, Goldie DJ. Change of insulin dosage, circulating free and bound insulin and insulin antibodies on transferring diabetics from conventional to highly purified porcine insulin. Diabetologia 1978; 14: 99-105.