Note: Descriptions are shown in the official language in which they were submitted.
35~
SYST~ FOR D~AND-BASED ADMINISTRATI~ ~F INSULIN
DESCRIPTION
Technical Field
This invention relates to means for ascer-
taining insulin demand of a patient based uponmeasurable changes of a body fluid physical property
and to means for controlled administration vf insulin.
`.
Background of Invention
Diabetes mellitus is a disease characteri~ed
by hyperglycemia, polyuria, and wasting. Hypergly-
cemia is due to decreased utilization of glucose andalso increased production of glucose.
The discovery of insulin in 1922 has made it
possible to control the blood glucose level in
!~ diabetic patients, at least partially. This enhanced
the well ~eing and survival of diabetic patients;
however, recent studies indicate that long term
complications of diabetes such as blindness, heart
failure, kidney failure, neuropathy and vascular
disease are not completely obviated by insulin
therapy. It has been suggested that the foreyoing
long term complications are due to one or more o~ the
followi~g reasons~ the control of glucose levels
by periodic injection of insulin is imperfect and
better glucose control is important, (2) the control
of metabolic substances other than glucose is
important and necessary for the management of the
diabetic patient, and (3) the levels of hormones
other than insulin also should be adjustedl i.e., the
glucagon concentration should be reduced.
To improve the control vf glucose levels,
several methods have been suggested. Among such
,}.
.: ~ , . .. .
j 3 5 ~ ~3
~ 2--
methods are the "close controll' method whereby a
patient is hospitalized with attendant frequent
assays of blood sugars and frequent insulin
injections (e.g., before each meal). To effect
control, the individual or the clinical laboratory
has to perform frequent blood sugar analyses on a
regular basis.
A further extension of the aforementioned
treatment by "close control" is the use of a contin-
uous or constant rate infusion of insulin usin~ aninsulin dispenser for infusion of this hormone into
the patient. With such a machine, the rate of
insulin administration must be predetermined by a
physician, and further requires the maintenance of a
steady diet and a continued uniform sensitivity of
the patient to insulin.
It has been proposed to use a sensor respon-
sive to the patient's glucose level'for regulating
the rate of insulin adminstration by the
aforementioned insulin dispenser. However, all
glucose sensors proposed to date have been unstable
in vivo when used for time periods in excess of
several weeks and thus are not very practical. A
variety of problems have been encountered with such
sensors: (1) Sensors that rely on the oxidation of
glucose (with glucose oxidase) exhibit stability
problems due to the inherent instability or
inactivation of such enzymes, (2) sensors relying on
the direct oxidation of glucose (by means of
electrodes) encounter undesirable polarization
phenomena on the electrode surfaces, (3~ the in vivo
period of reliability of heretofore known implanted
glucose sensors is shortened by fibrous or fibrinous
encasement, This is so because all heretofore known
glucose sensors are rate dependent. That is,
~ ~;3~6
--3--
the glucose concentration in a patient's blood is
indicated by the reaction rate of qlucose at the
sensor. Glucose in the blood must diffuse to the
electrode or to the enzyme present in the sensor.
For reliable sensor signal output, a constant mass
transfer resistance of glucose to the sensing element
must be maintained. Progressive fibrous or fibrinous
encasement of the sensing element continuously alters
such resistance and requires frequent recali~ration
of the sensor~
Accordingly, there exists a pressing need
for a reliable means for administering insulin to a
diabetic patient utilizing a stable implantable
sensor.
SummarY of the Invention
It has now been discovered that the glucose
level and insulin need of a patient can be reliably
ascertained by measuring at substantially equilibrium
conditions certain changes that have occurred in a
physical property, i.e., a property not involving in
its manifQstation a chemical change, of a body fluid
of the patient. The requisite amount of insulin for
diabetes management dispensed in response to a signal
generated by such measurement. Determination of body
fluid osmolality i5 particularly suitable for this
purpose. Typical body fluids whose physical proper-
ties can be monitored for this purpose are blood,
peritoneal fluid, subcutaneous fluid, or the like.
According to the present invention, a system
foe administration of insulin responsive to a
;~ patient's insulin requirement includes an insulin
dispenser adapted for infusion of insulin into the
; patient, a transducer means responsive to a physical
property of a body fluid of the patient and generat
ing a signal having a magnitude which is a function
" 5 ~35~
--4--
of the aforementioned physical property of the body
fluid, and preferably a function of or proportional
to osmolality of the body fluid; and a dispenser
control mean~ operably associated with the transducer
means and with the insulin dispenser to receive the
signal and to dispense insulin in response to the
magnitude of the received signal. The insulin
dispenser can be adapted for implantation, for ion-
tophoresis, or for subcutaneous or transcutaneous
infusion, as desired. Where the insulin dispenser is
externa~ to the patient's body, insulin delivery to
the patient is effected via a cannula or by percuta-
neous catherization using a catheter made of a bio-
compatible material. Preferably the catheter also is
provided with a barrier against infection.
In some cases it may be desirable to be able
to measure the concentration of other blood constit-
uents such as sodium or potassium levels. The
concentration of such electrolytes can be determined
by implanting a secondary transducer which can be
designed to measure the conductivity or ion concen-
tration of the blood fluid.
A preferred transducer means for the purpose
of the present invention is an implantable blood or
tissue fluid osmolality sensor or detector, e.g., an
implantable osmometer that generates an electrical,
mechanical or telemetry signal which, in turn,
controls the operation of t'ne in~ulin dispenser.
Another embodiment would have a primary transducer
sensing osmolality and one or more secondary trans-
ducers sensing electrolytic conductivity or ion
; concentration. A preferred insulin dispenser is a
portable unit that can be worn by the patient or
implanted in the patient, and includes an insulin
reservoir, a pump means, and a switch means respon-
sive to the signal or signals from one or more
. ~ i 6
--5--
sensors or detectors and controlling the delivery of
insulin.
In another preferred embodiment, an
implanted sensor is used in conjunction with a dual
channel catheter and has a shape enabling the sensor
or sensors to be either (a) removed and replaced or
(b) removed, cleaned of fibrinous or other material,
and replaced through a channel in the catheter other
than the channel carrying insulin to the patient.
Insulin from the reservoir can be delivered
to the patient intraperitoneally, intravenously, or
- subcutaneously, or by any other convenient means as
desired.
Brief DescriPtion of the Drawinqs
In the drawings,
FIGURE 1 is graphical correlation of blood
glucose level and blood osmolality in normal and
dia~etic mammals;
: FIGURE 2 is a graphical representation of
the effect of endogenous insulin on various mammalian
blood parameters;
FIGURE 2A is a graphical comparison of
measured and calculated osmolality derived from the
data shown in FIGURE 2.
;~ 25 FIGURE 3 is a gxaphical representation of
the correlation of osmolality to glucose, Na, and K,
~ in a typical normal mammal;
. FIGURE 4 is a block diagram showing a system
: for controlled administration of insulin that
embodies the present invention;
~ FIGURE 5 is a schematic representationl
partly in section, of a vapor pressure osmometer;
; FIGURE 6 indicates the thermocouple tempera-
ture curve ~or the osmometer shown in FIGURE 5;
FIGURE 7 is a schematic representation of an
implantable oncotic pressure osmometer;
~,
`35~
FIGURE 8 is a schematic representation,
partly in section, of a freezing point os~ometer;
FIGURE 9 is a schematic representation,
partly in section, of a boiling point osmometer;
FIGURE 10 is a schematic YieW of one embod-
iment of the present insulin infusion system includ-
ing a supply reservoir, a micropump controlled by a
microprocessor, and a dual channel catheter assembly
for delivering insulin to the patient and providin~
signals to the microprocessor representing glucose
levels in the patient;
FIGURE 11 is an enlar~ed cross-section of
the catheter shown in FIGURE 10;
FIGURE 12 is an enlarged view of an exem-
plary replaceable sensor partly broken away to show
interior detail;
FIGU~E 13 is an enlarged, exploded fragmen-
tary section of the proximal end of sensor tube of
the catheter and the replaceable cap therefori
FIGURE 14 is a cross-section taken generally
along plane 14-14 of FIGURE 11 showing the catheter
sensor channel and insulin channel;
FIGURE 15 is similar to FIGURE 11, and is
: shown here to include a secondary sensor;
FIGURE 16 shows the addition of a semiper-
meable membrane to encase the sensors;
FIGURE 17 is an enlarged view of a conduc-
tivity sensor partly broken away to show interior
detail;
FIGURE 18 is an enlarged view of a ion
concentration sensor partly broken away to show
interior detail; and
FIGURE 19 is a block diagram showing a
: system for controlled administration of insulin using
two sensors in the body.
~ . ,
~3!i ~ (
--7--
Descrip~ion of Preferred_Embodiments
While the major effect of insulin in the
mammalian body is the lowering of the blood sugar
concentrationl ;nsulin can affect t'ne concentration
of a number of other body substances including
potassium, phosphates, hydrogen, ketone bodies such
as ~-hydroxybutyrate and aceto acetate, fatty acid
levels, sodium, and glycerol. In the diabetic state,
the presence of many unidentified compounds is also
noted, which compounds are not normal constituents of
the body but which are believed to be metabolized by
the body after insulin infusion. In the brain such
~idiogenic osmols" can accumulate to levels of up to
40 milliosmols; and can cause cerebral edema or coma
during treatment of diabetic ketoacidosis.
Osmolality is defined as the sum of the
concentration of all solutes in a solution. Its
~ units are "osmoles" or total moles of solute per
! kilogram of solvent. In the blood, the concentration
of all chemical body substances is reflected by blood
osmolality whether such substances can be chemically
identified or not. It has been found that in the
diabetic state, the effects of insulin infusion are
reflected by a measurable change in blood osmolality
and in osmolality of other body fluids. Moreover, it
has been found that the detectable change in blood
osmolality following exogenous and/or endogenous
introduction of insulin has a greater abs~lute value
than would be expected from the change in ~he blood
glucose concentration. Changes in other substances
such as fatty acids, glycerol, or "idiogenic osmols"
are dissernible by measurement of osmolality, even i,
such substances are not measured directly. Since the
level of such substances is affected by the insulin
level, variations in body insulin demand can be
`
3 ~
--8
readily detected and accommodated so as to maintain
blood os~olality within a desired range.
There may be some cases where it may be
desirable to measure more than just the osmolality
level. An increase in osmolality can someti~es be
caused by severe physical exertion, dehydration or
ingesting large amounts of certain electrolytes, e.g.
table salt. While osmolality shows a good correlation
with glucose levels, this correlation can sometimes be
affected by these causes. Since a severely reduced
level of water intake results in diminished water
supply in both the intercellular and extracellular
fluids, there is a resultant increase in the
concentration of all metabolic substances. Such an
increase can be detected by an increase in conduc-
tivity or ion concentration since the majority of
extracellular osmoles are electrolytes. As the
osmolality increases due to dehydration or ingestion
of salt, the concentration of electrolytes increases
raising the electrolytic conductivity and ion
concentration. No large increase in electrolytic
conductivity or ion concentration is caused by an
increase in glucose or other substances controlled by
insulin. By measuring the increase in conductivity or
ion concentration in conjunction with the osmolality,
e.g., ~y two separate sensing means, it is possible to
~Iscreen out" increases in osmolality not caused by
increases in glucose and thus avoid an injection of
insulin when it may be inappropriate. The outputs
from the separate sensors may be subtracted from one
another or otherwise processed, for example by using a
microprocessor device so as to arrive at a control
signal for dispensing insulin to the patient.
FIGURE 1 graphically illustrates the
relationship between blood glucose concentration and
S t ~
g
blood osmolality in normal and in diabetic animals
during loss of diabetic control over a time period of
one day. Dogs were the experimental animals, and
alloxan was used to induce diabetes. Intravenous
glucose infusion was used to increase blood glucose
levels. Insulin was omitted in the diabetic animal
for 24 hours so as to induce loss of diabetic control.
The data points were collected over a one day period
and marked changes in osmolality occurred after the
infusion of glucose. The slope of the early part of
blood osmolality increase is relatively steep and thus
provides a sensitive indication that an insulin
infusion is needed.
In FIGURE 1, it is seen that in a normal
animal osmolality shows a weak relationship to blood
glucose during the intravenous glucose infusion.
Other solutes must be appearing or disappearing, to
cause osmolar changes in the blood as opposed to
glucose. In a diabetic animal, on the other hand,
osmolality levels were already high, before glucose
addition, and increased further in proportion to
glucose levels between 200 and 400 mg~. The slope of
increase is approximately 20 mOsm/kg per 200 mg%
change in glucose concentration, approximately twice
that predicted by the molecular weight of glucose
alone. Accordingly, there are other solutes in the
serum that contribute to this increase.
~ he time course of these osmolality ch ang es
is depicted in FIGURES 2 and 2A where various blood
chemistries are depicted for one of the n~rmal dogs.
Samples were collected and insulin levels determined
as set forth in the Example hereinbelow. Appropriate
response of insulin to a glucose load is exhibited.
It is seen that sodium, potassium, blood urea nitro-
gen (BUN) and protein change little during a
~ ~3~
glucose load, but that osmolality shifts dramatically,first increasing, then decreasing. The swings in
osmolali~y are believed to be due to changes in
"unmeasured" or `'idiogenic" osmoles; these osmoles are
indicated by the "osmol gap" calculated by subtracting
out the effects of urea, Na, anionsl and glucose from
total osmolality. The osmolality changes are due to
glucose and to other molecules, as yet unidentified.
A decrease in osmolality to a level below normal is
due in part to these unidentified molecules. In a
normal animal, after a glucose level increase, the
pancreas functions to return osmolality to normal.
A sensor directly or indirectly responsive to
osmolality changes in the body at a substantially
equilibrium condition can be utilized to control the
infusion of insulin and to minimize the adverse
effects of diabetes. Such a sensor is sensitive to
the degree of elevation of a number of important
molecules besides glucose and a signal generated by
the sensor controls insulin infusion, with resultant
hypo-osmolality of the body stopping further dispens-
ing of insulin. For implantation in a patient the
sensor means should be compact and of relatively light
weight, and preferably of a shape enabling removal
through a catheter or cannula. As shown in FIGURES 1
and 3, osmolality is correlated with glucose, there-
fore control of osmolality also controls glucose.
FIGURE 4 schematically illustrates a closed
loop system for an effective diabetes control utiliz-
ing blood, osmolality for example, as tlle indicator ofinsulin demand. Osmolality of other body fluids is
also suitable for this purpose. Transducer ll, such
as an os~olality sensor, is introduced or implanted
into pati~nt 10 in any convenient manner, e.g., within
i35 ~6
patient's vascular space, subcutaneously, or
intraperitoneally, so as to be in contact with a body
fluid such as blood, ~ubcutaneous fluid, peritoneal
fluid, or the like. This transducer means can also be
applied to body surfaces, e.g., mucosal membranes, so
as to be in contact with extracellular fluid, if
desired. In any case, the magnitude of the signal
generated by the transducer is a function of the
osmolality of the body fluid in contact therewith and
is utilized to control insulin infusion from reservoir
13 via a catheter, cannula, or similar means utilizing
a control unit that receives the transducer signal by
electrical or mechanical means, by telemetry, or in
any other convenient manner as will be discussed in
detail hereinbelow. Thus, a reliable indication of
the patient's insulin demand is obtained and the
demand can be satisfied by infusion of the desired
amount of insulin with attendant control of a variety
of metabolic sustances including glucose.
Osmolality alone or in conjunction with
electrolytic conductivity or ion concentration may be
measured in a variety of body fluids. There is rapid
equilibrium between the intracell~lar, vascular, and
interstitial components with respect to osmolality,
electrolytic conductivity and ion concentration. The
interstitial components are those fluids which are
outside of cells and outside of the vasculature, and
include peritoneal, subcutaneous, salivary, spinal
fluids, and the like. A sensor for osmolality as well
; 30 as for conductivity and ion concentration can be
placed in contact with any of these.
The measurement of osmolality may be per-
formed in several ways. There are four "colligative
properties" of solutions, which change in proportion
; 35 to osmolality: 1) vapor pressure, 2) oncotic
3 ~ 6
-12-
pressure, 3) freezing point, and 4) boiling point. In
each instance, the property measured is a function of
the water concentration of the sample; water
concentratio~ decreases as solute concentration
increases. Accordingly, preferably transducer 11
utiliæes one or more of the aforementioned colliga-
tive properties to generate an output signal.
One of the most common methods of measure-
ment of osmolality uses vapor pressure, defined as the
pressure which water vapor exerts leaving the surface
of a fluid. In a closed chamber, this pressure
reaches equilibrium with the pressure of vapor in the
gas above the fluid, and this pressure is proportional
to the concentration of water in the vapor. This
concentration of water may be measured conveniently
by observation of the "dew point."
A typical instrument for this purpose is the
Wescor Vapor Pressure Osmometer. An implantable
osmometer 14 utilizing the same principle is schemat-
ically depicted in FIGURE 5. In this particularosmometer, a body fluid permeable membrane 15,
together with housing 16 define chamber 17. In the
top portion of the chamber 17 is a very small thermo-
couple 19. When equilibrium of vapor pressure of the
body fluid and gas is attained, the thermocouple i3
cooled several degrees by electric current (t~rough
the Peltier ef~ect~. The electric current is then
stopped, and the temperature of the thermocouple
rises to the "dew point" as water condenses on the
thermocouple. A typical time-temperature relation-
ship during this procedure is shown in FIGURE 6. The
"dew point" depression is the difference between
ambient temperature and the dew point. As the vapor
pressure decreases the dew point depression becomes
larger. Lower vapor pressure indicates, o course, a
-13~
lower concentration of water in the sample solution,
that isy a higher concentration of solute (higher
osmolality). Vapor pressure determination of
osmolality has been found to be exceedingly sensitive
and generally reliable method for measuring liquid
osmolality. osmometer 14 can be calibrated against a
known standard prior to implantation, and periodic-
ally after implantation by drawing an aliquot of the
patient's blood and determining the osmolality there-
of extracorporeally. Alternatively, a second osmom-
eter r similar to osmometer 14 bu~ with a hermetically-
sealed chamber containing a known gas-water vapor
mixture, can be implanted to serve as a periodic
cali~ration means.
Oncotic pressure is another possible method
for deteemination of osmolality. Oncotic pressure is
defined as the pressure exerted across a semi-
permeable membrane because of the presence o~ imper-
meable solutes. If a solute cannot pass through a
membrane, its concentration is different on both
sides of the membrane. There then exists a gradient
of water concentration across the membrane. Because
such membranes are usually permeable to water, there
is a transfer of water across such a membrane. Such
transfer will continue until pressure gradients occur
to cause an equal transfer of water in the opposite
direction.
~ n osmolality sensor utilizing oncotic
pressure is shown schematicall;~ in FIGURE 7. A semi-
permeable membrane 21~ impermeable to certain solutes~such as glucose), is mounted on a rigid support 23
with a tube 25 leading to a pxessure transducer 27
which can be a piezoelectric gauge or the like.
Semi-permeable membrane 21, made, for example, from
polysulfone film, and support 23 are positioned under
-14-
the patient's skin 31 embedded in subcutaneous tissue
32. ~he tube and support are filled with a solvent
for body fluid constituents, such as water for
example. Water moves across the membrane due to the
concentration differences of solutes in water~ and
continues to move until pressure gradients on both
sides of the membrane equilibrate. The pressure
gradient existing at any given time is measured util-
izing a pressure gauge or differential transducer.
Such pressure is proportional to the concentration of
non-permeant solutes (such as glucose and/or
"unmeasured osmoles") in the body fluid in contact
with membrane 21.
FIGURE 8 schematically depicts a freezing
point osmometer. Because of a variety of physical
interactions of solutes with solvents, the freezing
point of a solution decreases as its solute concen-
.ration increases. Thus, freezing point depression
may be used to indicate osmolality of a solution. A
body fluid sample is received in container 33 that is
equipped with cooling coil 35. The sample is "super
cooled," then made to freeze during agitation with
stirrer 39. The temperature of solidification is
determined using a thermocouple 37.
FIGURE 9 depicts yet another possibility for
osmotic pressure and thus osmolality measurement,
that of boiling point elevation. The device com-
prises vessel 41 equipped with heating coil 43 and
thermocouple 45. Since increasing solute concentra-
tion results in lower solvent concentration, boiling
occurs at a higher temperature. The temperature at
which a body fluid sample boils is measured with
thermocouple 45, the temperature elevation being
effected by heating coil 43.
The vapor pressure and oncotic pressure
3 ~ ~ ~
-15-
measurements in particular may be made easily and
accurately by placing the sensor in various body
fluids. One particular advantage is that water is
extremely diffusable, and will allow rapid equilib-
rium within body fluids, such as the peritoneum. Forbedfast patients the osmometer can be a separate,
free-standing unit operably associated with an
insulin dispensing device.
For any of these osmolality sensors, the
generated signal is ba~ed on an equilibrium condi-
tion, i.e./ the signal is not dependent on the rate
at which a physical change takes place at the trans-
ducer, but rather on the equilibrium condition that
is encountered~ The problems encountered by prior
art sensors, due to fibrous or fibrinous deposits on
the sensor or transducerl are thus obviated or at
least minimized and stable, accurate readings are
obtained.
An increase in osmolality caused by dehydra-
tion or salt ingestion can also be detected by anelectrolyte concentration increase by measuring
electrolytic conductivity or ion concentration
measurements. In the case of electrolytic conduc-
tivity, the measurement can be made by placing within
the body biologically inert electrodes, such as
platinum, in a fixed geometric relation and measuring
the resistance value across the electrodes. Improved
measurements can also be made by the application of
alternating current at high frequency across the
electrodes. 10l000-20,000 Hertz is a useful fre-
~uency for this purpose.
Electrolytic conductivity can also be
measured by means of electrical induction without the
use of contacting electrodes. Such measurements are
made by inducing a current in the body fluid by use
; 3 ~ ~ ~
-16
of a coil of wire. The magnitude of the induced
current which can be measured by a second coil is
proportional to the conductivity of the body fluid.
Instead of or in addition to conductivity, it is also
possihle to measure the ion concentration of the
fluid.
Ion concentration can be determined by
measuring the electromotive force (voltage) between
two electrodes placed in the body fluid where one of
the electrodes is surrounded by a membrane chosen by
one knowledgeable in the art according to the body
fluid constituent wished to be measured. Such
measuréments could include total electrolytes,
particular electrolytes such as potassium or sodium,
pH, and disolved gases. In situations where there
may be possible temperature variations, the measuring
transducers could also include temperature
compensators.
The osmolality, electrolytic conductivity
and ion concentration sensors generate signals which
are transmitted by means of an appropriate lead or
leads, a radio signal, or similar expedients, to
dispenser control means 12 which, in turn, energizes,
or de-energizes, insulin dispenser 13, as indicated
for transcutaneous delivery of the requisite dose of
insulin. If a separate lead or leads to the sensor
are used, the conductor portion of the lead can be
coated with an inert, biocompatihle sheath.
Optionally, a fibrous cuff, e.g., 3 Dacron felt or
the like, can be provided around the biocompatible
sheath so as to form a barrier against infections.
Dispenser control means 12 can be a micro-processor,
- a relay network, or any other switching means adapted
to respond to the signal emitted by transducer means
11 and capable of energi~ing insulin dispenser 13.
~3
-17-
To ~he extent that fibrous or fibrinous deposits on a
transducer may hinder the attainment of an
equilibrium condition, control means 12 can include a
delay means that permits energization of insulin
dispenser 13 after a predetermined time period from
the point in time when the signal from transducer
means 11 is received. In this manner, consistent
actuation of dispenser 13 can be assured as long as
equilibrium can be attained at transducer means 11
within a predetermined time period. Alternatively,
control means 12 can include a timing device that
samples and compares signals received at predeter-
mined intervals and energizes dispenser 13 only after
differences among a plurality of received consecutive
lS signals fall within a predetermined range.
Insulin dispenser 13 includes an insulin
pool or reservoir and a pump means energizable from
any convenient power source, e.g., a primary or
secondary battery or gas propellant, in response to
an output signal received from dispenser control 12.
Instead of an osmotic pressure measurement r
the osmolality of the body fluid can also be deter-
mined using any other colligative property of the
body fluid. For example, the transducer means can be
adapted to measure freezing point depression, vapor
pressure, or boiling point evaluation.
Likewise, control of the desired dose deli-
very of insulin can be effected based on electrical
i~pedance measurements performed on the body fluid by
means of implanted inert electrodes, e.g., ~latinum,
; and an alternating current generator. Electrical
conductance measurements can also be used to obtain a
signal suitable for controlling the infusion of
insulin by means of dispenser 13.
In a method aspect of the present invention
a body fluid of the patient is contacted with a
. :
.
~ ~ G 3 ~
-18-
transducer means that is responsive to a physical
property of the body fluid, e.g., blood, which
property is indicative of the patient's insulin
deiciency, ~or example, osmolality or one of the
colligative properties thereof such as vapor pressure
or osmotic pressure. The transducer means i5 of the
type that generates an output signal which is a
function of the aforementioned physical property.
The magnitude of the output signal generated as a
consequence of the transducer mean~ contactin~ tlle
body fluid can then be utilized as an indicator of
the patient's insulin re~uirement as well as to
control the amount of insulin dispensed. Additional
measure~ents of other body fluid properties such as
electrolyte level, pH or dissolved gases by means of
electrolytic conductivity or ion concentration
measurements can be made, and the values thereof
utilized in conjunction with the obtained osmolality
value to provide a control signal t~ dispense the
required amount of insulin. A microprocessor is well
suited for this purpose.
The present invention is further illustrated
by the fol1owing example.
EX~MPLE
Materials and Methods
Animals: Five mongrel dogs were used. ~wo
were diabetic and three were non-diabetic. All dogs
except one had a permanent indwelling catheter, the
tip of which was in the cranial vena cava at the
level of the second intercostal space. The canula
exited from the jugular vein in the middle of the
neck and was tunneled under the skin to the withers
where it emerged. The free end was taped to a light
harness which the dog wore all the time. One dog 17
had a temporary jugular catheter implanted before the
IV glucose tolerance test.
5 ~ ~
19-
Diabetes Induction: Diabetes was induced in
two dogs (No. 17 and No. 22) with alloxan, 65 mg/kg.
Dog No. 17 had been diabetic for 4 years. The other
dog, No. 22, had been diabetic for 4 months. Insulin
was withheld from tlle diabetic dogs on the day of the
tests.
Glucose Tolerance Tests: Intravenous
glucose toler-lnce tests were performed twice on each
dog with the exception of the dog who had been
diabetic for 4 years. Four fasting heparinized blood
samples were drawn at 30 minute intervals to estab-
lish baseline values. A bolus of 50% glucose
(2 ml/kg body weight) was injected via the catheter.
Blood samples were drawn at 15 minute intervals for
90 minutes and then at 30 minute intervals for one
hour. Diabetic dogs were not given insulin on the
day of the glucose tolerance test.
Blood samples were centrifuged in a refrig-
erated centrifuge at 2000G's at 5C. for 15 minutes.
Plasma was removed, ali~uoted for subsequent tests
and frozen.
Blood glucose was measured by the glucose
oxidase-periodase method, Boehringer Mannheim Corp.,
Catalog No. 124036. Insulin was measured using the
Beckton Dickenson insulin assay kit, Catalog No.
231517. Insulin antibodies were removed from the
plasma of diabetic dogs prior to insulin assay
according to the method of Nakagawa et al., ~iabetes
22:590-600 (1973). Osmolality was measured using the
Wescor vapor pressure osmometer, Model No. 5130.
Sodium and potassium were analyzed by flame
photometry~ BUN was analyzed on the Beckman ~UN
Analyzer II. Protein was determined using Folin
Phenol reagent by the method of Lowrey et al.,
J. Biol. Chem. 193:265-275 (1951).
i 3 5 ~ ~;
-20-
Data Analysis: The mean and standard error
of the mean for each variable were calculated for
each time period during the IV glucose tolerance
test, for the normal and diabetic dogs. The data
were normalized by setting the mean baseline value
for each variable to zero in each test and calcu-
lating the difference between each observation and
its corresponding mean baseline value. One way
analysis of variance was performed on each of the
normalized variables over time to determine signif-
icance. Student-Newman-Reuls tests were performed to
make multiple comparisons. A Student's t-t2st was
used to compare the postinfusion osmole drop with
baseline data. Forwar3 least squares regression
analysis was used to estimate the relationship of
variables and to develop an equation for osmolality
incorporating all measured variables. A stepwise
regression analysis was used to pick the best set of
variables to predict osmolality. A calculated value
for osmolality was obtained using measurements of
BUN, sodium and glucose by the formula provided by
Edelman, J.S., ~eibman, J., O'Meara, M.P., and
Birkenfeld, L.W.: "Interrelationships between serum
sodium concentration, serum osmolality and total
exchangeable sodium, total exchangeable potassium and
total body water." J. Clin. Invest. 37: 1236-1256,
`~ 1958:
Osmolality - 1075 NA +0.0556 Glucose +0.357
BUN + 10.1
The use of this value is discussed in further detail
below.
Regression analyses of variance and
Student-Newman-Keuls tests were performed on the
Control Data Corporation's CDC 6500 computer using
the Regression and One-way Programs of the
Statistical Package for the Social Sciences (SPSS).
~ ~3S ~ ~ (
Results
Normal Dogs: Figure 2 shows the variation
in several blood parameters of one normal dog (No.
27), during ~n IV glucose tolerance test. During tl-e
1-1/2 hour baseline period the glucose, insulin,
sodium, potassium, BUN and protein remained
relatively constant. Osmolality varied within a 10
mOsmole range from 290 to 300 mOsmoles/kg, After
glucose infusion, blood glucose rose from 118 to 3S2
mg/dl. Insulin rose from a mean baseline value of
4UU/ml to 25g ~U/ml in 15 minutes and returned to the
baseline by 45 minutes. Osmolality rose from the
mean baseline value of 235 mOsmoles~kg to 310
mOsmoles/kg in 15 minutes. Over the next 30 minutes
the osmolality dropped by 25 mOsmoles/kg to 285
mOsmoles/kg.
In the aforementioned normal dog the corre-
lation ~Ir~ between osmolality and blood glucose was
0.80 (FIGURE 3). However t changes in blood glucose
were not sufficient to explain the osmolality changes
as the slope of this curve was greater than 1
mOsmole/ 180mg/L. The sodium remained relatively
constant during the entire test and showed no
correlation with osmolality, r = 0.00. Potassium
changes were inversely proportional t~ both osmotic
pressure (r = 0.64) and glucose changes (r = 0.62).
BUN and protein levels remained relatively constant
and showed little correlation with osmotic changes
(r - 0.25).
The mean values for the IV glucose tolerance
tests on all of the dogs are illustrated in FIGURE
2. During the baseline studies, the ranges in the
measured variables for normal dogs were glucose:
105-110 mg/dl; osmolali~y 286-290 mOsmoles/kg;
insulin; 17-32 ~U/ml; sodium: 147-150 mEq/l; X:
,..,
.
3 ~
-22-
4.1-4.4 mEq/l; BUN: 10-12 mg/dl, plasma protein:
5.4-5.9 g/dl. After intraveneous glucose adminis-
tration, glucose rose to 230 mg/dl. This represents
a 12.7 mOsmoles/kg contribution to osmolality. How-
ever, during this time the mean measured osmolalityrose only 9 mOsmoles/kg, and then fell to 7
mOsmolesfkg below the mean baseline level.
The maximum decrease in measured osmolali~y
for each dog occurred from 60 to 120 minutes after
glucose administration and is therefore obscured in
the mean data. The magnitude of the decrease ranged
from 2 to 18 mOsmoles/kg below the baseline. The
mean decrease in measured osmolality after intra-
venous administration of glucose was 10 + 3
mOsmoles/kg below baseline. A Student's t-test
showed this drop to be signiEicant at the ~= .05
level: t3_of5 - 2.352, calculated t = 2.94. This
decrease of osmolality below baseline cannot be
explained by a drop in any specific solute below
- 20 baseline levels.
Analysis of variancè showed no significant
diff~rences over time in mean sodium, BUN, and
protein levels after glucose infusion. The mean
potassium level dropped slightly at 30 minutes,
probably in response to the increase in insulin. The
potassium level then rose and was significantly
(p = 0.05~ higher from 120 to 150 minutes.
Diabetic Dogs: FIGURE 1 also shows the
variation in several blood parameters for one
diAhetic dog ~No. 22) during an IV glucose tolerance
test. During the 1 1/2 hour baseline period the blood
~lucose fell spontaneously from 544 to 436 mg/dli
~easured osmolali'y varied between 306 and 310
mOsmoles/~g; sodium rose from 126.1 to 138.0 mEq/l
during the baseline period. Insuli~, potassi~m and
,
3 5 ~ ~>
--23--
protein xemained relatively constant both during the
baseline period and after the infusion of glucose.
After glucose infusion the blood glucose rose by 294
mg/dl. In 2 1/2 hours glucose returned to the initial
level of 440 mg/dl. After the glucose infusion, the
measured osmolality rose only 5 mOsmoles/kg above the
baseline value at 15 minutes. Osmolality continued to
- rise at 30 minutes to 315 mOsmoles/kg even as the
glucose level was decreasing. The measured osmolality
then fell to a minimum of 304 mOsmoles/kg at 60
- minutes. It then rose to a maximum of 322 mOsmoles/kg
at 2 1/2 hours. Sodium decreased 3 mOsmoles after
glucose infusion, returned to the zero time level by
90 minutes, and then decreased again.
In this particular test the correlation
between blood glucose and osmolality was very low:
r = 0.03. The correlation between blood glucose and
osmolality in other IV glucose tolerance tests on
diabetic dogs were much higher: r = 0.95 and r = 0.75.
; 20 In this test on dog No. 22 there was a slight corre-
lation between sodium and osmolality (r - 0.24).
Osmolality was inversely proportional to potassium
(r = 0.36), ~UN (r = 0.38), and protein (r = 0.30),
however, these correlations are low.
The mean values of the measured plasma
constituents of all diabetic animals during IV glu-
; cose tolerance tests are also shown in FIGURE 2. Over
the baseline period, the values were: glucose, 418-439
mg/dl; sodium, 134-137 mEq/l; potassium, 4.0-4.2
mEq/l; BUN 13-15 mg/dl; protein, 5.6-5.9 g/dl. The
measured osmolality over the baseline period was
297-302 mOsmoles/kg.
After glucose infusion, in diabetic dogs the
mean blood glucose rose to 810 mg/dl and returned to
424 mg/dl by 2 1/2 hours. The insulin level remained
35~ (
-24-
relatively unchanged throu~hout the test. Sodium
levels decreased to 133 mEq/1 at 15 minutes and then
increased above baseline levels reaching a maximum of
142 mEq/l at 2 1/2 hours. Potassium levels remained
within the baseline range except for a 3.9 mEq/l value
- at 75 minutes. BUN decreased slightly showing a
minimum value of 12 mg/dl at 90 minutes. Protein also
showed a slight tendency to decrease after glucose
- infusion, reaching a minimum value at 75 minutes.
The increase in blood glucose after glucose
infusion represents a 21.2 mOsmoles/kg contribution to
osmolality. However, the mean measured osm~lality
after glucose infusion rose only 16 mOsmoles/kg above
the average baseline value. It then dropped to a
minimum of 301 mOsmoles/kg at 75 minutes and subse-
quently rose again to 309 mOsmoles~kg at 150 minutes.
After the initial rise following glucose
infusion osmolality did not decrease below the base-
line value in the long-term diabetic dog (NoO 17). In
the other diabetic dog (No. 223 the decrease was 3-6
mOsmoles/kg below baseline between 60 and ~5 minutes.
This decrease is not significantt however (p >.05).
In practicing the present invention, insulin
- can be infused as required utilizing a semi-permanently
implanted percutaneous cat'neter, preferably of the
type provided with a subcutaneously positioned fibrous
cuff, made of polyester felt or similar material, that
permits the ingrowth of tissue and capillary blood
vessels therein. In this manner the catheter is not
only fixed within the patient but also a barrier
against bacterial invasion is effectlvely maintained.
~. i
; !
$~
.
3 5 ~ ~ ~
-2~
i
. .
The continuous infusion of insulin suhcutan-
eously, intraperitonially or into a vein in response
to a change in the physical property of a body fluid
provides a more effective control of the patient's
blood sugar level than is currently possible. More-
over, through the continuous infusion of insulin, t~e
rate of insulin absorption is not influenced by such
factors as exercise and temperature. In addition to
the continuous delivery of insulin to the patient as
controlled by transducer or sensor 11, the system
shown in the aforementioned copending application
permits the patient to elect to have an additional
infusion (bolus) as required, for example, at or
shortly before a meal.
The system includes a reservoir and a pump
j pack adapted to be strapped to the torso of a patient
at an implanted catheter. The catheter is adapted to
extend under the skin down the front of the chest witl
the ti? near the entrance to the heart in a central
vein.
This system includes a wearable pack includ-
ing a one-piece prefilled insulin reservoir bag. When
the insulin supply becomes exhausted and must be
replaced, the pump segment (the heavy tube portion
that is engaged by rollers in the pump) is removed by
; the roller section of the pump and the pump segment of
a new supply is threaded through the pump there~y
eliminating the pump as a contaminant to the system.
The delivery from the pump is controlled by a
microprocessor programmed to respond to an input or
inputs from the implanted osmolality sensor and
possi~ly secondary sensors and is also designed to
~, .
P~35 ~6
-26-
permit the patient to anticipate the need for an extra
quantity (bolus) of insulin. In the latter instance a
push-button control on the side of the wearable pack
can be actuated for the bolus infusion. The micro-
processor can also be programmed to prevent thepatient from activating the bolus injection more than
a predetermined number of times a day, depending upon
the physician's prescription which can be prese~ in
the microprocessor.
In another system embodyin~ the present
invention a reservoir and pump pack are adapted to be
strapped to the torso of the patient. A dual channel
catheter is adapted to extend under the skin ~subcu-
taneously) down under the chest with the tip of the
catheter near the entrance of a central vein. Other
methods of vascular access may be employed depending
on the requirements of the patient. Likewise, sub-
cutaneous or peritoneal infusion of insulin may be
effected. The catheter carries a cuff or sleeve of
"Dacron" or other material into which the patient's
tissue and blood vessels grow for permanent i~planta
tion, thereby reducing or obviating the possibility of
bacterial infection.
The catheter itself is designed for prolonged
implantation and can be constructed of a flexible
silicone rubber (Silastic)*or other physiologicall~
compatible material having two parallel channels
therethrough. One of these channels is the insulin
infusion channel that is externally connected to the
insulin pump. The other channel receives a lead wire
or wires for the osmolality sensor and possible
secondary sensors which project from and are posi-
tioned by the distal end of the sensor channel.
~n important as~pect of the present invention
is that the dual channel catheter permits the removal
*Trade Mark
.~
3~
-27-
and replacement (or removal, cleaning and replacement)
of the osmolality sensor and any secondary sensor
without the removal of the catheter itself. ~ecause
of fibrous or fibrinous deposits or the general
degradation of the transducer means or sensors after a
. prolonged period of use it is necessary that they be
periodically replaced or cleaned to prevent it
adversely affecting the insulin delivery function.
Toward this end, the external end of the catheter is
split defining an insulin tube connectable to the
insulin pump and a sensor tube having a releasable and
removable cap through which sensor lead wires project.
This cap has a conical projection that fits in and
over and seals the end of the sensor tube. The cap
also has a central bore therethrough that sealingly
receives the sensor lead wires.
After a certain period of use, for example
six months, a sensor can be replaced or cleaned, if
desired, by removin~ the cap and withdrawing the
sensor through the. sensor channel and thereafter
replacing the cleaned sensor or replacing it with a
new sensor and cap. The position of the cap on the
sensor lead wire determines the extent of projection
and positioning of the sensor itself from the distal
~5 end of the sensor channel.
: Referring to FIGURE 10, the insuli.n infusion
system utilizin~ a dual lumen tchannel) catheter is
seen to include a replaceable insulin supply 51, small
roller pump 52 controlled by a microprocessor 53 and
an implanted catheter 54 connected to tl~e insulin
supply 51 through a releasable connector assembly 56.
In lieu of pump 52 other types of small pumps can be
used, for example a piezoelectric micropump of the
general type shown in ~.S. Patent No. 3,963,380 and
the like.
& 3 ~
-28-
WhilP not shown in FIGURE 10, the insulin
supply 51~ small pump 52 g and microprocessor 53 can be
assembled into a single pack adapted to be worn by the
patient in any convenient manner, e.g., strapped to
the torso, or, if sufficiently small, adapted for
implantation.
Insulin supply 51 is a one piece plastic
insert and includes a polyethylene insulin reservoir
bag 57, tube section 58, pump segment 59, and outlet
tube 60 and supply connector 61, the latter forming
part of the connector assembly 56~ The p~ump segment
59 of the insulin supply can be a segment of tube
section 58 or a relatively heavy-walled section 67
terminàting in pump blocks 65 and 66. Heavy walled
section 67 is adapted to be engaged by the rollers 55
of the micropump 52. When the insulin supply 51 is
replaced, the patient discards the exha~sted supply
unit, threads the pump section 59 of a new supply
through the pump 52, and connects connector 61 to the
connector assembly 56.
The pump 52 can be a relatively small roller
pum~. The microprocessor 51 can be a digital logic
system appropriately programmed for controlling the
infusion rate of pump 5~ in accordance with signals
provided by osmolality sensor 70 carried intracorpor
eally by the catheter 54 and providing signals ~o the
- microprocessor through a lead wire 71. The details of
the necessary logic in the microprocessor 53 are
readily apparent to an electronics engineer ?f ordinary
skill given the intended function so the detailed
3~6
-29-
schematic therefor is not included in the drawings.
Suffice it to say that the microprocessor 53 includes a
variable pulse generator for driving stepper motor 61
associated with pump 52 at a variable rate. An input
circuit in microprocessor 53 responsive to changes in
signal levels in sensor lead wire 71 biases the vari-
able pulse generator and will increase the pulse rate
to the stepper motor 61 in response to increases in the
insulin demand, e.g., glucose level as sensed by sensor
70 in the patient's circulatory system and decreases
the pulse rate to the pump 52 in response to decreased
insulin demand sensed by sensor 70. In ~his manner
insulin infusion rate is increased as patient's
osmolality and glucose levels increase, and insulin
infusion rate is decreased in response to decreased
osmolality and/or glucose levels in a substantially
continuous manner throughout the day.
The microprocessor 51 can also be progra~med
to provide automatically a higher insulin infusion rate
at predetermined time periods durin~ a 24-hour cycle,
or the microprocessor can be pr~grammed to make and
store a series of insulin demand determinations at
predetermined intervals, to extrapolate therefrom an
anticipated peak demand, and to control insulin
infusion rate accordingly,
As seen in greater detail in FIGURE 11,
catheter 54 includes a left external section, an
intermediate subcutaneous section A and an intraven-
tricular section B. A tissue-impregnable cuff 72
surrounds the catheter 54 at the juncture of the
external and subcutaneous sections to provide the
catheter during prolonged implantation with an excel-
lent bacteriological barrier. The catheter 54 in-
cludes an insulin passage 73 having a connector 74 at
the end thereof adapted to be connected to connector 56
~ ~ ~ 3 5 s ~
-30-
to receive insulin from pump 52. Passage 73 has a
relatively narrow section 75 at its distal end that
extends intravascularly.
: The catheter also includes a smaller dia-
meter sensor channel 77 that receives the replaceable
sensor lead wire 71 and positions the sensor 70 adja-
cent its intravascular distal end 79. The distal end
73 of t'ne catheter passage is spaced a considerable
distance from the distal end of the insulin passage 75.
The external end of the catheter 54 is splity
forming an insulin tube 80 and a sensor tube 81. A
replaceable cap 82 is provided for the sensor tube 81
to seal the sensor and also to permit the re~oval and
cleaning or removal and replacement of the sensor 53
periodically to prevent fibrous or fibrinous buiId up
or other degradation of the sensor 70 from adversely
affecting the response of the microprocessor 53.
- As illustrated in FIGURE 13, the proximal end
of tube 81 has a plurality of annular integral projec-
tions 83 that hold and form a labyrinth seal with
corresponding annular recesses 84 in the interior bore
85 of cap 82. To further seal sensor passage 77 from
contamination, cap 82 has a narrow central opening 87
therethrough that sealingly receives lead wire 71 from
sensor 70. The cap 82 is positioned on the lead wire
71 at a distance so that when the sensor is replaced,
the sensor head 70 will be properly positioned the
desired distance from the distal end 79 of the sensor
: passage 77 within the patient's blood vessel.
When periodic replacement of tne sensor 70 is
required, cap 82 is removed and the lead wire 71
withdrawn, withdrawing sensor 70 through sensor pas-
sage 77. A temporary cap may be attached to the
catheter to prevent infection in the interior. There-
after the cleaned sensor or a new sensor 70 is
.' ' ' .
.~ .
' ~3~
-31-
inserted, with the new cap 82, and the cap replaced
connected to tube 81 and to the correct position of cap
82 on wire 71. The sensor 70 is then properly
positioned.
The osmolality sensor 70 as shown in FIGURE 12
is a vapor pressure osmometer of the type illustrated
in FIGURE 5 and comprises a pair of matched thermo-
couples 91 and 93 situated in respective chambers 95
and 97. Chamber 95 is a hermetically sealed enclosure
containing water vapor in equilibrium with liquid water
at body temperatures. Chamber 97 is substantially the
same as chamber 95 but for the fact that a wall portion
98 thereof is made of a semi-permeable membrane such as
polysulfone film, cellulose acetate film, or the like,
so as to permit e~uilibration of water vapor pressure
within chamber 97 with that of the surrounding body
fluid. Standard dialysis membranes that prevent the
passage of the relatively smaller solute molecules but
that keep out proteins are also suitable for this
purpose. To minimize accumulation within the chambers
between measurements, ~he chambers can be pressurized
to drive out the substances contained therein and to
permit a new equilibrium to be established prior to
making the next measurement A sterile dry air or
gaseous nitrogen sweep of the chambers can also be
utilized for this purpose. Thermocouple 91 in chamber
95 provides a reference value for dew point of water at
the body temperature e~isting at the time the
osmolality measurement is made. For ascertaining
sodium ion concentration in the body fluid,
particularly suitable is a glucose-impermeable, sodium
ion-permeable cellulose acetate membrane commercially
avail-ble from Osmonics Corporation, Hopkins,
Minnesota, under the designation Sepa-2.
The use of two sensing tran`sucers is shown in
-32-
FIGURE 15 which is a modification of FIGURE 11. The
osmolality sensor 70 and its lead 71 are as before,
added are a secondary sensor 100 and its lead 99. The
choice of this secondary sensor can be for conductivity
or ion concentration.
The addition of semipermeable membrane capsule
101 which can be placed around the sensors shown in
FIGURE 16. This capsule can be attached to the sensors
and be withdrawn with them. In a preferred embodiment
the semipermeable membrane w~uld be impermeable to
bacteria but permeable to ~lucose and electrolytes and
would form a seal about the distal end of the catheter
79, thus providing a second barrier against infection.
A cutaway view of the conductivity is shown in
FIGURE 17. The sensor has an open end 104 in the
casing 103 and the ~wo electrodes 102. Fluid enters
the casing, and measurements are made by measuring
resistance across the two electrodes.
In FIGURE 18 is shown a cutaway view of an
electromotive force detector with its casing 107, an
open end 108, the exposed electrode 106, the sealed
electrode 109 and the membrane 105. The membrane is
chosen by one knowledgeable in the art depending on the
body fluid constituent wished to be measured. This
includes pH, total ion concentration or the
concentration of partic~lar ions such as potassium or
sodium.
FIGURE 19 is a modification of FIGURE 4 to
show the use of a secondary transducer 110 placed in
the body 10 to provide an additional input for the
microprocessor 12 which controls delivery of insulin
from the reservoir 130
The foregoing discussion and the accompany-
ing drawings are intended as illustrative, and are not
to be taken as limiting. Still other variations and
rearrangements of parts within the spirit and scope of
. ,
~, ....
~3
-33
the present invention are pos6ible and will readily
present themselves to one skilled in the art.
1 0
- 15
i
'
: 25
, ~
~ ..,
,
., ,
