Mechanism of Awareness of

Hypoglycemia
Perception of Neurogenic (Predominantly
Cholinergic) Rather Than Neuroglycopenic
Symptoms
DWIGHT A. TOWLER, CAROLYN E. HAVLIN, SUZANNE CRAFT, AND PHILIP CRYER

We sought 1) to determine which symptoms of

hypoglycemia are reproducible, 2) to
pharmacologically distinguish neurogenic (autonomic)
from neuroglycopenic symptoms, and 3) to test the
hypothesis that awareness of hypoglycemia is the
result of perception of neurogenic rather than
neuroglycopenic symptoms. Awareness of
hypoglycemia and 19 symptoms were quantitated in 10
normal, young adults, each studied on four occasions
in random sequence, during 1) clamped euglycemia
(~5 mM), 2) clamped hypoglycemia (-2.5 mM), 3)
clamped hypoglycemia with combined a- and
p-adrenergic blockade (phentolamine and propranolol),
and 4) clamped hypoglycemia with panautonomic
blockade (phentolamine, propranolol and atropine).
Significant (ANOVA, P < 0.001) treatment effects on
the awareness of hypoglycemia ("blood sugar low")
were noted. No change occurred in the score for this
during euglycemia, but the mean SE increase was
2.1 0.4 during hypoglycemia. This increase was not
reduced significantly by adrenergic blockade
(1.6 0.5), but was reduced significantly and
substantially (-70%) by panautonomic blockade
(0.6 0.3). Significant neurogenic symptoms included
shaky/tremulous ( P < 0.001), heart pounding
(P < 0.001), and nervous/anxious (P = 0.002), all
adrenergic; and sweaty (P < 0.001), hungry (P < 0.001),
and tingling (P = 0.009), all cholinergic. Significant

From the Division of Endocrinology, Diabetes, and Metabolism of the Department of Medicine; and the General Clinical Research Center; and the
Diabetes Research and Training Center, Washington University School of
Medicine, St. Louis, Missouri.
Address correspondence and reprint requests to Dr. Philip E. Cryer,
Division of Endocrinology, Diabetes, and Metabolism (Box 8127), Washington
University School of Medicine, 660 South Euclid Avenue, St. Louis, MO 63110
Received for publication 12 May 1993 and accepted in revised form 26 July
1993.
IDDM, insulin-dependent diabetes mellitus; RIA, radioimmunoassay; EPI,
epinephrine; NE, norepinephrine; NEFA, nonesterified fatty acid; HR, heart
rate; BP, blood pressure; sBP, systolic blood pressure; dBP, diastolic blood
pressure; ANOVA, analysis of variance; BMI, body mass index; (S-OHB,
p-hydroxybutyrate.

DIABETES, VOL. 42, DECEMBER 1993

neuroglycopenic symptoms, those produced by

hypoglycemia but not reduced by panautonomic
blockade, included warm (P < 0.001), weak (P = 0.011),
difficulty thinking/confused (P = 0.004), and
tired/drowsy (P = 0.003). We conclude that muscarinic
cholinergic mechanisms mediate an important and
previously uncharacterized component of the
neurogenic symptoms of hypoglycemia and awareness
of hypoglycemia. Awareness of hypoglycemia is
largely, perhaps exclusively, the result of perception of
neurogenic rather than neuroglycopenic symptoms.
Diabetes 42:1791-98, 1993

atrogenic hypoglycemia is a major problem for patients with IDDM because of intermittent hyperinsulinemia that results from imperfect insulin replacement
regimens and compromised physiological defenses
against hyperinsulinemia (1). In many patients, this problem is compounded by the syndrome of hypoglycemia
unawareness. In this syndrome, the patient experiences
partial or complete loss of the warning symptoms of
developing hypoglycemia that previously prompted its
recognition and led to food ingestion, which prevented
hypoglycemia's progression to severe hypoglycemia ( 1 7).
Conventionally, symptoms of hypoglycemia are divided into those directly attributable to glucose deprivation from the CNS (neuroglycopenic symptoms) and
those attributable to the autonomic nervous system activation triggered by hypoglycemia (neurogenic or autonomic symptoms) (8). Although some symptoms are
readily placed in one category or the other, others are not
(9-11). Clinical hypoglycemia unawareness in IDDM is
often assumed to be the result of the loss of neurogenic
symptoms of hypoglycemia (2,4), but this has not been
documented (8). Indeed, patients recall neuroglycopenic
as well as neurogenic symptoms after an episode of
hypoglycemia (10,11). Therefore, it is conceivable that

1791

AWARENESS OF HYPOGLYCEMIA

(45 mg/dl) through 120 min. On one of these three

occasions,
saline was infused from 0 through 120 min.
2. HYPOGLYCEMIA
-5.0 mmol/L-H
2.5mmol/L*
On
another
occasion, the nonselective a-adrenergic
-Solineantagonist
phentolamine
mesylate (Regitine, Ciba3. HYPOGLYCEMIA +
-5.0 mmol/ L-H
2.5 mmol/LGeigy,
Summit,
NJ)
and
the
nonselective (3-adrenergic
ADRENERGIC
-Phentolamine + Propranolol(70pg/kg70
(14 g/lg*1.4
BLOCKADE
antagonist
propranolol
hydrochloride
(Inderal, Ayerst,
' '
'
New York, NY) were infused from 0 through 120 min. And
4. HYPOGLYCEMIA* h
-5.0 mmol/L
-H2.5mmol/Lon another occasion, phentolamine and propranolol were
PANAUTONOMIC
I
Phentolamine + Propranololinfused from 0 through 120 min, with the muscarinic
BLOCKADE

Atropine
(l.Omg)
cholinergic antagonist atropine sulfate (Elkins-Sinn,
Cherry Hill, NJ) injected intravenously at 0 and 60 min.
SEGMENTS:
AI
-BThe doses of these antagonists and the experimental
-60
-30
0
30
60
design are shown in Fig. 1.
TIME (minute.)
Observations (arterialized venous blood samples, BPs,
FIG. 1. Experimental design.
HRs, and symptom scores) were made at 15-min intervals (-30 through 120 min). Plasma glucose concentrations were, of course, measured more frequently.
patients use both neuroglycopenic and neurogenic Cognitive function tests were performed between 45 and
symptoms to recognize developing hypoglycemia. Thus, 60 min and 105 and 120 min.
whether awareness of hypoglycemia is normally based Analytical methods. Plasma glucose concentrations
on the perception of neurogenic symptoms, neuroglyco- were determined with a glucose oxidase method on an
penic symptoms, or both is not known. Likewise, whether analyzer (Beckman, Fullerton, CA). Plasma insulin (13),
clinical hypoglycemia unawareness in IDDM is the result C-peptide (13), glucagon (14), pancreatic polypeptide
of loss of neurogenic symptoms, neuroglycopenic symp- (15), growth hormone (16), and cortisol (17) concentratoms, or both is unknown.
tions were determined with RIA. Plasma EPI and NE
To determine which of an array of reported symptoms concentrations were assessed with a single isotope
are reproducible for hypoglycemia and which are neuro- derivative (radioenzymatic) method (18). Enzymatic
genic as opposed to neuroglycopenic in origin and to methods were used to measure serum NEFA (19) and
test the hypothesis that awareness of hypoglycemia is blood p-OHB (20), lactate (21), and alanine (22) concenthe result of perception of neurogenic rather than neuro- trations. BPs and HRs were recorded automatically (Dyglycopenic symptoms, we quantitated these variables namap, Critikon, Tampa, FL).
during euglycemia, hypoglycemia alone, hypoglycemia
Subjects scored each symptom from 0 (none) to 6
with combined a- and p-adrenergic blockade, and hy(severe). The symptoms included 16 that were thought to
poglycemia with panautonomic (adrenergic and muscabe indicative of hypoglycemia, i.e., difficulty thinking/
rinic cholinergic) blockade in normal, young adults.
confused, warm, shaky/tremulous, nausea, tired/drowsy,
hungry, weak, sweaty, headache, heart pounding, diffiRESEARCH DESIGN AND METHODS
culty speaking, nervous/anxious, dizzy, faint, tingling,
We studied 10 normal humans (7 men, 3 women), 22-29
yr of age with BMIs ranging from 18.7 to 30.7 kg/m2. and blurred vision; and three not thought to be indicative
Subjects gave their written consent to participate in this of hypoglycemia, i.e., cold, happy, sad. "Blood sugar
study, which was approved by the Washington University low" was used as a measure of awareness of hypoglySchool of Medicine Human Studies Committee and con- cemia.
Cognitive function tests included measures of global
ducted at the Washington University General Clinical
Research Center. The subjects were given no instruc- cognitive function (line orientation), attention (vigilance
tions about the symptoms of hypoglycemia before the and serial addition), and memory (immediate and delayed paragraph recall), as described previously (23).
study. However, they all had medical backgrounds.
Experimental design. As illustrated in Fig. 1, each Statistical analysis. To assess the impact of the various
subject was studied (after an overnight fast) on four antagonist drugs used, the mean incremental change in
occasions, using a hyperinsulinemic euglycemic and each parameter from segment A (mean of -30, -15, and
hypoglycemic clamp technique (12), in random se- 0 min values) to segment B (mean of 30, 45, and 60 min
quence. On all occasions, regular human insulin (Humu- values) (Fig. 1) under all four study conditions was tested
lin, Lilly, Indianapolis, IN) was infused intravenously in a with a repeated measures ANOVA. To assess the impact
dose of 12.0 pmol kg" 1 min ~1 (2.0 mU kg" 1 of hypoglycemia, the mean incremental change in each
min ~1) and plasma glucose concentrations were parameter from segment B to segment C (mean of 90,
clamped at 5.0 mM (90 mg/dl) with variable infusions of 105, and 120 min values) under all four study conditions
20% glucose before initiation of observations (at - 3 0 also was tested with a repeated measures ANOVA. For
min). On one occasion, euglycemia (-5.0 mM) was those parameters with significant overall treatment efmaintained through 120 min. On the other three occafects, differences among individual conditions were
sions, euglycemia was maintained through 60 min; then
tested with the Newman-Keuls multiple-range test. Data
glucose levels were lowered and clamped at 2.5 mM
are presented as means SE.
1. EUGLYCEMIA

5.0 mmol / L-

1792

DIABETES, VOL. 42, DECEMBER 1993

D.A. TOWLER AND ASSOCIATES

6.0

100
5.0

4.0

80

GLUCOSE

o
1

3.0

60

m-- EU
= HYPO

40

o= HYPO+ADB

2.0

a = HYPO + PAB
-30

30

60

90

120

GLUCAGON

o
I 0.25

20

30

75

150
_j

100

50

30

60

90

^
a

60

90

120

PANCREATIC
POLYPEPTIDE

300

50

25

200

$rt
-30

120

30

90

120

TIME (min)
FIG. 2. Mean SE plasma glucose, Insulin, C-peptlde, glucagon, and pancreatic polypeptlde concentrations during hyperinsuiinemic (12.0
pmol kg~ 1 mirr 1 ) euglycemlc (-5.0 mM) clamps {EU) (-30 to 120 min) and euglycemic then hypoglycemic (-2.5 mM) clamps (60-120 mln)
alone {HYPO), with combined oc and p-adrenergic blockade {HYPO + ADB, phentolamlne and propranolol 0-120 min) and with panautonomic
blockade {HYPO + PAB, phentolamine and propranolol 0-120 mln plus atropine at 0 and 60 min).

RESULTS
Hormonal, metabolic, and hemodynamic changes.
Plasma glucose targets were achieved (Fig. 2). Plasma
insulin concentrations were comparable under all study
conditions. Plasma C-peptide levels were unaffected
by the antagonists; they fell during hypoglycemia
(P = 0.003) but were unaffected by adrenergic or panautonomic blockade (Fig. 2). Plasma glucagon and pancreatic polypeptide concentrations were unaffected by the
antagonists during euglycemia. The glucagon responses
to hypoglycemia (P = 0.020) were unaffected by adrenergic or panautonomic blockade (Fig. 2). The pancreatic
polypeptide responses to hypoglycemia (P< 0.0001)
were unaffected by adrenergic blockade, but were completely prevented by panautonomic blockade (Fig. 2).
Plasma EPI concentrations were unaffected, but plasma
NE levels were increased (P< 0.0001) by the antagonists during euglycemia. Both the EPI (P< 0.0001) and
NE (P< 0.0001) responses to hypoglycemia were increased by both adrenergic and panautonomic blockade
(Fig. 3). Plasma growth hormone and cortisol concentrations and their responses to hypoglycemia (both
P < 0.0001) were unaffected by adrenergic or panautonomic blockade (Fig. 3).
Serum NEFA and blood lactate, p-OHB and alanine
concentrations were unaffected by the antagonists during euglycemia. The NEFA (P = 0.002) and lactate

DIABETES, VOL. 42, DECEMBER 1993

(P< 0.0001) responses to hypoglycemia were prevented by both adrenergic and panautonomic blockade
(Fig- 4).
HRs were unaffected by adrenergic blockade but were
increased (P < 0.0001) by panautonomic blockade (Fig.
5) during euglycemia. sBPs decreased (P = 0.031) more
prominently during adrenergic compared with panautonomic blockade during euglycemia; dBPs were unaffected (Fig. 5). Mean BP (data not shown) also
decreased ( P = 0.003). The HR responses to hypoglycemia (P< 0.0001) were prevented by both adrenergic
and panautonomic blockade (Fig. 5). Apparent differences in sBPs during hypoglycemia (Fig. 5) were not
statistically significant.
Awareness of hypoglycemia. No significant effects of
the antagonists (phentolamine and propranolol with or
without atropine) on the scores for "blood sugar low"
were noted during euglycemia (segment A vs. segment
B) (data not shown). However, significant (P< 0.001)
treatment effects were observed during hypoglycemia.
No change was observed in the mean SE score from
euglycemia to euglycemia, but the score increased significantly (2.0 0.4) from euglycemia (segment B) to
hypoglycemia (segment C) (Fig. 6). The score during
hypoglycemia was not significantly reduced by adrenergic blockade (1.6 0.5), but it was reduced significantly

1793

AWARENESS OF HYPOGLYCEMIA
1
1

EPI15.0 . NEPHRINE

- 4000

12.0

- 3000

90

20.0

2000

NOREPINEPHRINE

1500

i
i

r?n ; EU
10.0

5.0
0

- 2000 a

= HYPO
o = HYPO + ADB
D= HYPO+PAB
-30

30

//

^
- 1000

60

90

500

3.0

120

-30

30

60

90

120

80

GROWTH
HORMONE

3000

1000 a

I 6.0

60

2000
40

1000
20

-30

30

60

90

120

TIME (min)
FIG. 3. Mean SE plasma EPI, NE, growth hormone, and cortlsol concentrations during hyperlnsullnemic euglycemic clamps (EU) and
euglycemic then hypoglycemic clamps alone (HYPO), with combined a- and p-adrenergic blockade (HYPO + ADB) and with panautonomic
blockade (HYPO + PAB).

and substantially (70%) by panautonomic blockade

(0.6 0.3) (Fig. 6).
Symptoms. No significant effects of the antagonists on
the scores for any symptoms were noted during euglycemia (segment A vs. segment B) (data not shown).
However, significant treatment effects on the scores for
several of the symptoms were observed during hypoglycemia (segment B vs. segment C) (Figs. 6-8).
Scores for six symptoms, i.e., sweaty (P< 0.001),
hungry (P< 0.001), tingling (P= 0.009), shaky/tremulous (P< 0.001), heart pounding (P< 0.001), and nervous/anxious (P= 0.002), increased significantly during
hypoglycemia; and these increases were reduced or
prevented by adrenergic blockage, panautonomic
blockade, or both (Fig. 6). The increments in scores for
sweaty, hungry, and tingling during hypoglycemia were
reduced by panautonomic but not adrenergic blockade.
The increments in scores for shaky/tremulous, heart
pounding, and nervous/anxious were reduced by both
adrenergic and panautonomic blockade.
Scores for four symptoms, i.e., warm (P< 0.001), weak
(P = 0.011), difficulty thinking/confused (P = 0.004), and
tired/drowsy (P= 0.003), increased significantly during
hypoglycemia, but these increases were not reduced by
adrenergic or panautonomic blockade (Fig. 7). Scores
for four other symptoms, i.e., faint, dizzy, difficulty speaking, and blurred vision, appeared to increase during

1794

hypoglycemia and to be unaffected by adrenergic or

panautonomic blockade (Fig. 7). However, changes in
scores for these symptoms did not achieve overall statistical significance.
Scores for five symptoms, including headache and
nausea and the three not thought to be symptoms of
hypoglycemia (cold, sad, and happy), did not change
significantly during hypoglycemia (Fig. 8).
Cognitive function. The antagonists had no significant
effect on the line orientation task during hypoglycemia,
but significant effects were found on the vigilance
(P < 0.01), serial addition (P < 0.001), immediate verbatim paragraph recall (P< 0.04), and delayed verbatim
paragraph recall (P< 0.001) tasks (Table 1). The increase in the number of vigilance errors tended to be
greater during hypoglycemia with adrenergic blockade
and was significantly greater during hypoglycemia with
panautonomic blockade. The increase in the number of
serial addition errors was greater during hypoglycemia
with adrenergic and panautonomic blockade. The decrease in the number of informational bits recalled immediately in the verbatim paragraph-recall task was
significantly greater during adrenergic blockade and
tended to be greater during hypoglycemia with panautonomic blockade. The decrease in the number of bits
recalled later was greater during hypoglycemia with
adrenergic and panautonomic blockade.

DIABETES, VOL. 42, DECEMBER 1993

D.A. TOWLER AND ASSOCIATES

200

200

NEFA

3200-

- 3200

150

I 100

E23= E U

= HYPO
50

o = HYPO + ADB
o = HYPO + PAB
-30

30

60

90

120

-30

400 -

30

60

90

- 200

200

_,

200 -

-30

30

60

120 '

90

" -30

30

TIME (min)
FIG. 4. Mean SE serum NEFA and blood Iactate, aianine and p-OHB concentrations during hyperinsulinemic euglycemic clamps (EU) and
euglycemic then hypoglycemic clamps alone (HYPO), with combined a- and p-adrenergic blockade (HYPO + ADB) and with panautonomic
blockade (HYPO + PAB).

DISCUSSION
These data indicate that awareness of hypoglycemia
normally is largely, if not exclusively, the result of perception of neurogenic rather than neuroglycopenic symptoms; this perception is mediated predominantly by
muscarinic cholinergic rather than adrenergic mechanisms. When asked to score the likelihood that their blood
glucose was low, subjects were clearly able to distinguish euglycemia (-5.0 mM) from hypoglycemia (~2.5
mM). However, subjects were significantly and substantially less able to distinguish euglycemia from hypoglycemia during panautonomic blockade (with phento-

30

30

60

90

120

TIME (min)
FIG. 5. Mean SE HRs, sBPs, and dBPs during hyperinsulinemic
euglycemic clamps (EU) and euglycemic then hypoglycemic clamps
alone (HYPO), with combined a- and p-adrenergic blockade
(HYPO + ADB) and with panautonomic blockade (HYPO + PAB).

DIABETES, VOL. 42, DECEMBER 1993

lamine, propranolol, and atropine). Changes in scores

were reduced from 2.1 0.4 during hypoglycemia alone
to 0.6 0.3 during hypoglycemia with panautonomic
blockade. Notably, the subjects were also able to distinguish euglycemia from hypoglycemia during adrenergic
blockade. Changes in scores during hypoglycemia with
adrenergic blockade (1.6 0.5) were not significantly
different from those during hypoglycemia alone. Clearly,
therefore, although a minor adrenergic component cannot be excluded, awareness of hypoglycemia is largely
mediated by atropine-sensitive (muscarinic cholinergic)
mechanisms. The extent to which this is mediated by
sympathetic or parasympathetic cholinergic postganglionic neurons cannot be determined from this data. In
either event, because panautonomic blockade reduced
awareness substantially, the data indicate that awareness of hypoglycemia normally is largely the result of
perception of neurogenic rather than neuroglycopenic
symptoms. Stated differently, even though both neurogenic and neuroglycopenic symptoms are perceived, it
is the neurogenic symptoms that are interpreted to indicate the presence of hypoglycemia. Thus, it is reasonable to attribute clinical hypoglycemia unawareness in
IDDM (1-7) to loss of neurogenic symptoms of developing hypoglycemia.
Although reduced substantially, awareness of hypoglycemia was not prevented completely by panautonomic blockade. This may indicate that neuroglycopenia
also plays some role in awareness of hypoglycemia.

1795

AWARENESS OF HYPOGLYCEMIA
SWEATY (p<0.00l)

SHAKY/TREMULOUS

* t

I
8*
to

HEART POUNDING

SYMPTOMS

(p< 0.001)

i
ADB

PAB

0
BLOOD SUGAR LOW
|p< 0.001)

NERVOUS/ANXIOUS

0
EU

ADB

PAB

FIG. 6. Mean SE changes (A) in symptom scores from segment B to

segment C (Fig. 1) during euglycemia ((/, ED), hypoglycemia alone
(HYPO, ) , hypoglycemia plus combined adrenergic blockade
(HYPO + ADB, ^ ) , and hypoglycemia plus panautonomic blockade
(HYPO + PAB, E H ) for neurogenic (autonomic) symptoms of sweaty,
hungry, tingling, shaky/tremulous, heart pounding, and
nervous/anxious and the recognition of low blood glucose.
*, Significant differences between the data shown in adjacent columns;
t , significant differences between HYPO and HYPO + PAB.

Alternatively, it may be that neurogenic responses were

not blocked completely with the competitive antagonists
used. Although substantial adrenergic and cholinergic
blockade was produced, as discussed later, blockade
may not have been absolute.
These data also permit identification and classification
of many of the symptoms of hypoglycemia. Of the 19
symptoms (in addition to "blood sugar low") studied, 10
were significantly increased by hypoglycemia. Because
they were either reduced substantially or prevented
entirely by panautonomic blockade, six of the symptoms
TIRED/DROWSY
|p 0.003)

NEUROGLYCOPENIC SYMPTOMS
DIFF. SPEAKING

WEAK (P-0.011)

(p-0.310)

Iii
DIFF. THINKING/
CONFUSED

BLURRED VISION

DIZZY (p=0.130)

|p = 0.161)

(p 0.004)

ADB
-HYPO

PAB

AOB
-HYPO

PAB

EU

ADB

PAB

-HYPO

FIG. 7. Mean SE changes (A) in symptom scores from segment B to

segment C (Fig.1) during euglycemia (EU, ED), hypoglycemia alone
(HYPO, I B ) , hypoglycemia plus combined adrenergic blockade
(HYPO + ADB, &Z2), and hypoglycemia plus panautonomic blockade
(HYPO + PAB,) for neuroglycopenic symptoms of warm, weak,
difficulty thinking/confused, tired/drowsy, faint, dizzy, difficulty
speaking, and blurred vision. *, Significant differences between the
data shown in adjacent columns.

1796

HAPPY (p0.697)

EU

AOB PAB
-HYPO

EU

ADB PAB
-HYPO

0ADB

PAB

-HYPO

FIG. 8. Mean SE changes (A) In symptoms scores from segment B

to segment C (Fig. 1) during euglycemia (EU, ED), hypoglycemia alone
(HYPO, ) , hypoglycemia plus combined adrenergic blockade
(HYPO + ADB, ^ ) , and hypoglycemia plus panautonomic blockade
(HYPO + ADB, E H ) for symptoms of cold, headache, nausea, sad,
and happy, which were not found to be produced by hypoglycemia.

HYPO

HYPO

WARM (P<0.00l)

SAD (p= 0.135)

< 2

(p= 0.002)

(p=0.071)

TINGLING (P-0.009)

EU

HEADACHE

NONSYMPTOMS

NEUROGENIC

HUNGRY lp<0.00i)

V'<<-\

NAUSEA (p = 0.056)

El

1.

COLD (p=0.185)

| p < 0.001)

studied (i.e., shaky/tremulous, heart pounding, nervous/

anxious, sweaty, hungry, and tingling) are neurogenic in
origin. The symptoms of shaky/tremulous, heart pounding and nervous/anxious were reduced substantially by
adrenergic blockade alone and, therefore, are at least
predominantly adrenergic neurogenic symptoms. The
extent to which these symptoms are mediated by EPI
released into the circulation from the adrenal medullae,
NE released into synaptic clefts from terminals of sympathetic postganglionic neurons in relevant target tissues, or both cannot be determined from these data. In
contrast, the symptoms of sweaty, hungry, and tingling
were not reduced by adrenergic blockade alone but
were reduced by panautonomic blockade. Thus, these
are at least predominantly cholinergic neurogenic symptoms. Alternatively, four of the symptoms studied (i.e.,
warm, weak, difficulty thinking/confused, and tired/
drowsy) were significantly increased by hypoglycemia
but were not reduced by panautonomic (or adrenergic)
blockade. Thus, these are most reasonably viewed as
neuroglycopenic symptoms. Inspection of the data suggests that the symptoms of faint, dizzy, difficulty speaking, and blurred vision may also fit in this category, but
the changes in the scores for these did not achieve
overall statistical significance. Although we think it unlikely, it is conceivable that one or more of the symptoms
considered neuroglycopenic might be mediated by a
nonadrenergic, noncholinergic neural mechanism. Finally, neither the symptoms of headache and nausea nor
the nonspecific symptoms (cold, happy, and sad) were
found to be reproducible symtoms of hypoglycemia.
Several approaches have been taken to classify symptoms of hypoglycemia. One approach is to deduce the
symptom category from a known mechanism of the
physiological response thought to underlie that symptom
(9). For example, it is well established that the diaphoretic
response to hypoglycemia is cholinergic (9,24), evidence
supports that the increase in finger tremor during hypoglycemia is adrenergic (9,25), and it has been reported
that palpitations do not occur during hypoglycemia in
persons with adrenal medullary denervation (26). Thus,
perhaps it is not surprising that we find the symptom of

DIABETES, VOL. 42, DECEMBER 1993

DA. TOWLER AND ASSOCIATES

TABLE 1
Cognitive function test scores during euglycemia, hypoglycemia alone, hypoglycemia with adrenergic blockade and
hypoglycemia with panautomatic blockade
Nominal plasma glucose (mM)

Test
Line orientation (errors)
Vigilance (errors)
Serial addition (errors)
Immediate recall (bits)
Delayed recall (bits)

Overall
P
value

Saline
5.0

NS
0.3
0
<0.01
<0.001 0.7
<0.04 14.0
<0.001 13.4

5.0
0.2

Adrenergic
blockade

Saline

0.3
0.3
0.3 0.5
1.4 15.4
1.4 14.8

5.0

2.5

0.2 0.3
0.2 0.2
0.2 1.0
1.4 12.6
1.2 11.4

0.2 0.3
0.1 0.7
0.3 0.9
1.3 13.6
1.3 11.8

5.0

2.5

0.2 0.6 0.3

0.6
0.2 0.4 0.2
1.5
0.3 0.5 0.2* 2.2
1.4 13.7 1 . 5 * 11.2 +
1.5 11.8 1 . 1 t 8.9

Panautomatic
blockade
5.0
0.3 0.2 0.2
0.5 0.1 0 . 1 *
0.6 0.4 0 . 2 t
1.4 14.0 1.6
1.4 12.0 1.6*

2.5
0.3
2.1
3.6
12.2
9.0

0.2
0.6
0.6
2.0
1.7

Data are means SE.

*P < 0.01 for the change compared with that during euglycemia throughout,
t P < 0.001 for the change compared with that during euglycemia throughout.
$P < 0.05 for the change compared with that during euglycemia throughout.

sweaty to be cholinergic and those of shaky/tremulous

and heart pounding to be adrenergic. However, this
approach is not applicable to the other symptoms of
hypoglycemia, including hungry and tingling, which we
find to be cholinergic neurogenic, and nervous/anxious,
which we find to be adrenergic neurogenic.
Another approach, used by Hepburn etal. (10,11), is to
analyze symptoms from a large number of hypoglycemic
episodes by factor analysis so that symptom clusters can
be identified. Considerable agreement exists between
the conclusions drawn from that approach and our
present findings; however, some differences exist. Hepburn et al. placed the symptom of hungry in the nonspecific (10) or the neurogenic category (11). We find it to be
neurogenic. They considered the symptom of warm to be
neurogenic (10). We find it to be neuroglycopenic. They
considered the symptom of weak to be neuroglycopenic
(10) or nonspecific (11). We find it to be neuroglycopenic.
Our interpretation that a reproducible symptom of
hypoglycemia that is not reduced by panautonomic
blockade is a neuroglycopenic symptom rests on the
premise that a high level of a- and p-adrenergic and
muscarinic cholinergic blockade was, in fact, produced.
Evidence of effective autonomic blockade before hypoglycemia includes 1) a decrease in sBP (a, p), 2) an
increase in plasma NE concentrations (a) (27,28), and 3)
an increase in HR (muscarinic). Evidence of effective
autonomic blockade during hypoglycemia includes 1) no
increase in HR (3), 2) greater increases in plasma EPI
and NE concentrations (P) (29,30), 3) no increase in
serum NEFAand blood lactate concentrations (p), and 4)
no increase in plasma pancreatic polypeptide concentrations (muscarinic) (31). Thus, substantial hemodynamic and neuroendocrine evidence exists that a high
level of adrenergic and cholinergic blockade was produced.
Based on these data, a reasonable panel for the
quantitation of symptoms of hypoglycemia could include
the adrenergic neurogenic symptoms of shaky/tremulous, heart pounding, and nervous/anxious; the cholinergic neurogenic symptoms of sweaty, hungry, and
tingling (as well as "blood sugar low"); and the neuro-

DIABETES, VOL. 42, DECEMBER 1993

glycopenic symptoms of warm, weak, difficulty thinking/

confused, and tired/drowsy.
Interestingly, the antagonists altered cognitive function
during hypoglycemia. They did not affect line orientation,
a task that measures the integrity of visuoperceptual
processes with only minimal attentional demands (23).
Thus, no gross global disturbance of cognitive function
occurred. On the other hand, attention, as assessed by
the vigilance and serial addition tasks, was impaired by
adrenergic blockade and to a greater extent by panautonomic blockade during hypoglycemia. The latter implies involvement of cholinergic as well as adrenergic
pathways. This impairment of attention may well explain
the poorer memory performance, which was assessed by
the immediate and especially the delayed verbatim paragraph recall tasks, during hypoglycemia with adrenergic
and panautonomic blockade. Nonetheless, the fact that
attention and memory were impaired during hypoglycemia is potentially relevant to patients with IDDM treated
with medications that have anti-adrenergic or anti-cholinergic actions.
In summary, muscarinic cholinergic mechanisms mediate a previously unrecognized large proportion of neurogenic symptoms of hypoglycemia (sweaty, hungry,
tingling) and mediate awareness of hypoglycemia. Adrenergic mechanisms mediate other neurogenic symptoms (shaky/tremulous, heart pounding, anxious/
nervous). Awareness of hypoglycemia is largely, if not
exclusively, the result of perception and interpretation of
neurogenic rather than neuroglycopenic symptoms.

ACKNOWLEDGMENTS

This study was supported in part by United States Public

Health Service Grants DK-27085, DK-44235, DK-20579,
DK-07120,and RR-00036.
The authors acknowledge the technical assistance of
Suresh Shah, Krishan Jethi, Gregory Winter, Shirley Hill,
and Joy Brothers; the assistance of the nursing staff of
the Washington University General Clinical Research
Center, particularly Virginia Bischof, in the performance
of these studies; the data management assistance of

1797

AWARENESS OF HYPOGLYCEMIA

Daniel Flasar; the statistical advice of Dr. Curtis Parvin;

and the assistance of Mary Russo in the preparation of
the manuscript.

REFERENCES
1. Cryer PE: latrogenic hypoglycemia in IDDM: Consequences, risk
factors and prevention. In Diabetes Annual. Vol. 7. Marshall S, Home
PD, Alberti KGMM, Krall L, Eds. Amsterdam, Elsevier. 1993, p.
317-31
2. Pramming S, Thorsteinsson B, Bendtson I, Binder C: Symptomatic
hypoglycemia in 411 type I diabetic patients. Diabetic Med 8:21722, 1991
3. Ryder REJ, Owens DR, Hayes TM, Ghatei M, Bloom SR: Unawareness of hypoglycemia and inadequate glucose counterregulation:
no causal relationship with diabetic autonomic neuropathy. Br Med
J 301:783-87, 1990
4. Heller SR, Herbert M, Macdonald IA, Tattersall RB: Influences of
sympathetic nervous system on hypoglycemic warning symptoms.
Lancet ii:35963, 1987
5. Hepburn DA, Patrick AW, Eadington DW, Ewing DJ, Frier BM:
Unawareness of hypoglycaemia in insulin-treated diabetic patients:
prevalence and relationship to autonomic neuropathy. Diabetic Med
7:711-17, 1990
6. Gerich JE, Korytokowski M, Mokan M, Veneman T, Mitrakou A:
Hypoglycemia unawareness. Endocrinol Rev 12:356-71, 1991
7. Cryer PE: latrogenic hypoglycemia as a cause of hypoglycemiaassociated autonomic failure in IDDM: a vicious cycle. Diabetes
41:255-60, 1992
8. Cryer PE, Binder C, Bolli GB, Cherrington AD, Gale EAM, Gerich
JE, Sherwin RS: Hypoglycemia in IDDM. Diabetes 38:1193-99,
1989
9. Heller SR, Macdonald IA: Physiological disturbances in hypoglycemia: Effect on subjective awareness. Clin Sci 81:1-9, 1991
10. Hepburn DA, Deary JJ, Frier BM, Patrick AM, Quinn JD, Fisher BM:
Symptoms of acute insulin-induced hypoglycemia in humans with
and without IDDM: factor analysis approach. Diabetes Care 14:
949-57, 1991
11. Hepburn DA, Deary JJ, Frier BM: Classification of symptoms of
hypoglycemia in insulin-treated diabetic patients using factor analysis: relationship to hypoglycaemia unawareness. Diabetic Med
9:70-75, 1992
12. Schwartz NS, Clutter WE, Shah SD, Cryer PE: The glycemic thresholds for activation of glucose counterregulatory systems are higher
than the threshold for symptoms. J Clin Invest 79:777-81, 1987
13. Kuzuya H, Blix PM, Horwitz DL, Steiner DF, Rubenstein AH: Determination of free and total insulin and C-peptide in insulin-treated
diabetics. Diabetes 26:22-29, 1977
14. Ensinck JW: Immunoassays for glucagon. In Glucagon: Handbook
of Experimental Pharmacology. Vol. 66. New York, Verlag, 1983
15. Gingerich RL, Lacy PE, Chance RE, Johnson MG: Regional pancre-

1798

16.
17.
18.

19.
20.
21.
22.
23.
24.
25.

26.
27.

28.
29.
30.
31.

atic polypeptide concentration and in vitro secretion of canine

pancreatic polypeptide, insulin and glucagon. Diabetes 27:96-101,
1978
Schalch D, Parker M: A sensitive double antibody radioimmunoassay for growth hormone in plasma. Nature 203:1141-42, 1964
Farmer RW, Pierce CE: Plasma cortisol determination: radioimmunoassay and competitive binding compared. Clin Chem 20:411-14,
1974
Shah SD, Clutter WE, Cryer PE: External and internal standards in
the single isotope derivative (radioenzymatic) assay of plasma
norepinephrine and epinephrine in normal humans and persons
with diabetes mellitus or chronic renal failure. J Lab Clin Med
106:624-29, 1985
Hosaka N, Kikuchi N, Mitsuhida N, Kawaguchi A: New calorimetric
methods for the determination of free fatty acids with acyl-CoA
synthetase and acyl-CoA oxidase. Biochem J 89:1799-1803, 1981
Pinter JK, Hayashi JA, Watson JA: Enzymatic assay of glycerol,
dihydroxyacetone and glyceraldehyde. Arch Biochem Biophys 121:
404-14, 1967
Lowry OH, Passoneau JV, Hasselberger FX, Schultz DV: Effect of
ischemia on known substrates and cofactors of the glycolytic
pathway of the brain. J Biol Chem 239:18-30, 1964
Cahill GF Jr, Herrera MG, Morgan AP, Soeldner JS, Steinke J, Levy
PL, Rerchand GA Jr, Kipnis DM: Hormone-fuel interrelationships
during fasting. J Clin Invest 45:1751-69, 1966
Hirsch IB, Boyle PJ, Craft S, Cryer PE: Higher glycemic thresholds
for symptoms during B-adrenergic blockade in IDDM. Diabetes
40:1177-86, 1991
Chalmers TM, Keele CH: The nervous and chemical control of
sweating. Br J Dermatol 64:43-54, 1952
Kerr D, Macdonald IA, Heller SA, Tattersall RB: A randomized,
double-blind placebo controlled trial of the effects of metoprolol CR,
atenolol, and propranolol LA and the physiological responses to
hypoglycaemia in nondiabetic subjects. Br J Clin Pharmacol 29:
685-94, 1990
French EB, Kilpatrick R: The role of adrenalin in hypoglycaemic
reactions in man. Clin Sci 14:639-51, 1955
Cryer PE, Haymond MW, Santiago JV, Shah SD: Norepinephrine
and epinephrine release and adrenergic mediation of smokingassociated hemodynamic and metabolic events. N Engl J Med
295:573-77, 1976
Tse TF, Clutter WE, Shah SD, Cryer PE: Mechanisms of postprandial
glucose counterregulation in man. J Clin Invest 72:278-86, 1983
Cryer PE, Rizza RA, Haymond MW, Gerich JE: Epinephrine and
norepinephrine are cleared through fj- adrenergic but not a-adrenergic mechanisms in man. Metabolism 29:1114-17, 1980
Best JC, Halter JB: Blood pressure and norepinephrine spill over
during propranolol infusion in humans. Am JPhysiol248:R400-406,
1985
Schwartz TW: Pancreatic polypeptide: a hormone under vagal
control. Gastroenterology 85:1411-25, 1983

DIABETES, VOL. 42, DECEMBER 1993