- Email: [email protected]
The Effects of Epidural Betamethasone on Blood Glucose in Patients with Diabetes Mellitus
Original Research
The Effects of Epidural Betamethasone on Blood Glucose in Patients with Diabetes Mellitus Peter Gonzalez, MD, Scott R. Laker, MD, William Sullivan, MD, Jeri E. F. Harwood, PhD, Venu Akuthota, MD Objective: To determine the effects of lumbosacral transforaminal and caudal epidural betamethasone injections on blood glucose levels in diabetic subjects. The hypothesis is that epidural steroid injections result in transient elevation of blood glucose levels in diabetic subjects. Design: This is a prospective, observational cohort. Twelve diabetic subjects (6 non– insulin-dependent and 6 insulin-dependent) receiving lumbosacral or caudal epidural betamethasone injections for neurogenic claudication or radicular pain were studied. Spinal level and approach were decided based on symptoms, pathology, and magnetic resonance imaging findings. Subjects recorded their finger stick blood glucose levels twice daily for 3 days before the injection, the day of the injection, and 3 days after the injection. Setting: A tertiary, university-based, spine center. Participants: Inclusion criteria included diabetic subjects (age 18 years) with the ability and willingness to monitor and report their blood glucose. Exclusion criteria included epidural steroid injections (ESIs) within the previous 2 months or peripheral corticosteroid injections within the previous 2 weeks. Nineteen subjects initially enrolled, and 12 successfully completed the study. Interventions: After informed consent was obtained, subjects underwent fluoroscopically guided lumbosacral transforaminal ESIs (TFESIs) or caudal ESIs, using contrast to confirm targeted needle placement and to rule out vascular uptake. Main Outcome Measures: Subjects recorded morning and evening blood glucose (mg/dL) via glucometer. Results: There was a 106 mg/dL average elevation in blood glucose level on the evening of the injection day. The blood glucose elevation remained statistically significant for 3 days after the injection (P ⬍ .002). Blood glucose remained elevated through postinjection day 2, though these elevations were not statistically significant. Conclusions: Lumbosacral transforaminal and caudal epidural betamethasone injections are associated with statistically significant elevations in blood glucose levels in diabetic subjects. This effect peaked on the day of the injection and lasted approximately 2 days.
INTRODUCTION Epidural steroid injections (ESIs) have been used in the treatment of lumbosacral radicular pain. They have been increasingly used in conjunction with conservative treatment, such as medications and therapeutic exercise to reduce pain in the short term and to obviate surgery [1]. It has been postulated that corticosteroids, such as betamethasone, decrease radicular pain through a reduction of the inflammatory cascade. Proposed mechanisms of action include decreased production of prostaglandins, histamine, leukotrienes, interleukin-1, interleukin-2, interferons, and tumor necrosis factor [2-5]. Hyperglycemia is a reported adverse effect of ESIs [6]. This effect is thought to be due to the opposition of corticosteroid to insulin with increased gluconeogenesis in the liver and decreased glucose uptake in the peripheral tissues [7]. Ward et al [8] reported on insulin sensitivity after ESIs. One previous study reported on glucose tolerance in 10 healthy, nondiabetic individuals receiving caudal ESIs (80 mg triamcinolone) and found that glucose and insulin levels significantly increased 24 hours after injection and returned to normal at PM&R
340
1934-1482/09/$36.00 Printed in U.S.A.
P.G. University of Colorado Denver, School of Medicine, Department of Physical Medicine and Rehabilitation, PO Box 6511, Mail Stop F-493, Aurora, CO 80045. Address correspondence to: P.G.; e-mail: peter.gonzalez@ ucdenver.edu Disclosure: nothing to disclose S.R.L. University of Washington, School of Medicine, Department of Rehabilitation Medicine, Seattle, WA Disclosure: nothing to disclose W.S. University of Colorado Denver, School of Medicine, Department of Physical Medicine and Rehabilitation, Aurora, CO Disclosure: nothing to disclose J.E.F.H. University of Colorado Denver, School of Medicine, Department of Pediatrics, Aurora, CO Disclosure: nothing to disclose V.A. University of Colorado Denver, School of Medicine, Department of Physical Medicine and Rehabilitation, Aurora, CO Disclosure: nothing to disclose Disclosure Key can be found on the Table of Contents and at www.pmrjournal.org Submitted for publication February 20, 2008; accepted December 17, 2008.
© 2009 by the American Academy of Physical Medicine and Rehabilitation Vol. 1, 340-345, April 2009 DOI: 10.1016/j.pmrj.2008.12.007
PM&R
1 week after injection [8]. Maillefert et al [9] reported on the systemic effects of epidural administration of 15 mg of dexamethasone acetate. These authors found that a single injection resulted in transient adrenal suppression indicating passage of corticosteroid into the bloodstream. Though there have been several studies that have reported on endocrinological sequelae of steroid injections, this is one of the first studies that reports directly on blood glucose levels in an exclusively diabetic population receiving lumbosacral transforaminal ESIs (TFESIs) and caudal ESIs. One case reported hypercorticism with Cushingoid features induced by 2 lumbar ESIs in a nondiabetic patient. The patient’s symptoms resolved over the course of 6 weeks [10]. Another study found that the hypothalamic-pituitary axis was altered for an average of 1 month after a series of 3 ESIs given weekly over 3 weeks [11]. The retrospective review by Botwin et al [6] of complications in lumbar ESIs found 1 patient that reported a transient elevation in blood sugar. Most recently, 1 study found that the corticotropic axis was altered up to 21 days after ESIs in both diabetic and nondiabetic patients [12]. Related research on the effects of injectable corticosteroids on hyperglycemia has revealed similar results. A study of patients receiving trigger finger injections with corticosteroid showed a statistically significant hyperglycemic effect on postinjection day 1 and a hyperglycemic trend that lasted 5 days [13]. A study of patients receiving daily periocular steroid injections found a peak in blood glucose 6 hours after the injection and a trough blood glucose level approximately 24 hours later that was 13% higher than baseline levels [14]. Younes et al also found that the corticotropic axis was altered in patient’s receiving intra-articular shoulder injections [12]. However, another study on the effects of intra-articular shoulder steroid injections showed no effect on blood glucose in diabetic patients [15]. The present study was designed to evaluate the effects of epidural betamethasone injections on blood glucose in subjects with diabetes mellitus receiving treatment for lumbar radicular pain. The intent was to quantify the effect on blood glucose and to determine the onset and duration of the changes. The hypothesis was that ESIs result in transient elevation of blood glucose in diabetic subjects.
MATERIALS AND METHODS This study was designed as a prospective, observational cohort that evaluated the effects of lumbosacral TFESIs and caudal ESIs on blood glucose in diabetic patients. The clinical investigators were not blinded given the observational design of the study. Randomization was not performed, because all subjects underwent the same types of interventions. Approval by the Multiple Institutional Review Board was obtained and recruitment of subjects was completed on May 1, 2007. Diabetic patients scheduled to undergo lumbosacral TFESIs or caudal ESIs were invited to participate in the study. All subjects were recruited at a university subspecialty spine
Vol. 1, Iss. 4, 2009
341
Figure 1. Flow diagram.
center setting. Informed consent was obtained by resident, fellow, or attending physicians. Inclusion criteria were subjects with an age greater than 18 years, a medical history of documented insulin-dependent diabetes mellitus (IDDM) or non–insulin-dependent diabetes mellitus (NIDDM), and the ability and willingness to record their blood glucose levels with a glucometer twice daily for 3 days before the injection, the day of the injection, and for 3 days after injection. This follow-up period was determined based on the results of previous studies [8,9,13,14]. Subjects that had received ESIs within the preceding 2 months and those that had received other joint or soft-tissue corticosteroid injections within the preceding 2 weeks were excluded. These time frames were based on previous related research [8-14]. The primary outcome measures were the blood glucose recordings from each subject’s home glucometer. All subjects were provided with a blood glucose log to monitor blood glucose levels each morning
342
Gonzalez et al
EPIDURAL STEROIDS AND EFFECTS ON BLOOD GLUCOSE
Table 1. Demographic data Age
Sex
Type
Diagnosis
Procedure
65 70
M F
IDDM NIDDM
B L5-S1 TFESI S1 TFESI
84
F
NIDDM
55 73 79
M F F
NIDDM IDDM NIDDM
60 77 55 67 87 62
M F F M M F
IDDM NIDDM IDDM IDDM NIDDM IDDM
Stenosis Stenosis with radiculopathy Stenosis with radiculopathy Radiculopathy Stenosis Stenosis with radiculopathy Radiculopathy Radiculopathy Stenosis Stenosis Stenosis Stenosis with radiculopathy
B L4-L5 TFESI L4-L5 TFESI L5-S1 TFESI L4-L5 TFESI L4-L5 TFESI L4-L5 TFESI Caudal ESI B TFESI Caudal ESI L4-L5 TFESI
Abbreviations: M ⫽ male; F ⫽ female; IDDM ⫽ insulin-dependent diabetes mellitus; NIDDM ⫽ non–insulin-dependent diabetes mellitus; TFESI ⫽ transforaminal epidural steroid injection; ESI ⫽ epidural steroid injection; L ⫽ lumbar; B ⫽ bilateral.
and evening. This log was returned to the investigators by mail or at subsequent follow-up appointments. The approach for administration of epidural corticosteroid (lumbosacral transforaminal or caudal approach) and level was decided based on symptoms, spinal pathology, and magnetic resonance imaging findings. All procedures were performed under fluoroscopic guidance by one of three treating investigators (P.G., W.J.S., V.A.). For both lumbosacral TFESIs and caudal ESIs, patients were placed in a prone position on an examination table within the fluoroscopy suite. The skin overlying the target level was cleansed with 10% povidone-iodine, USP prep solution (Novation Inc., Irving, TX). A solution of 1% lidocaine 10 mg/mL (Astra Zeneca; Wilmington, NC) was used to anesthetize the skin and soft tissues. All procedures were performed with 22-gauge, 3.5-inch spinal needles (Sherwood Medical, St. Louis, MO). Needle manipulation was performed under intermittent fluoroscopic guidance. Instillation of iohexol contrast material 240 mgI/mL (GE Healthcare, Oslo, Norway) was performed during each injection to insure proper placement of the needle tip within the epidural space and to confirm no vascular uptake of the injectate. A solution of 1-1.5 mL of 1% lidocaine was administrated in the epidural space prior to the administration of corticosteroid. The corticosteroid used for all injections was 6 mg/mL betamethasone (Schering Corp., Kenilsworth, NJ). Unilateral TFESIs consisted of 2 mL of betamethasone and bilateral injections used 1.5 mL of betamethasone per side, for a total of 3 mL. All caudal ESIs consisted of a mixture of 2 mL of betamethasone, 4 mL of normal saline, and 4 mL of 1% lidocaine.
Statistics All analyses assumed a 2-sided test of hypothesis and were performed in SAS v9.1 (SAS Institute, Inc., Cary, NC). A mixed-effects model was used to model glucose as a function
of day. These models accounted for the correlation between repeated measures on patients. Days ⫺3 through 0 include the 7 glucose measurements obtained preinjection and days 0.5 through 3.5 included all postinjection measurements. The model allowed a slope change at day 0 (ie, at the time of injection) and postinjection glucose was modeled as a function of polynomials of time, to appropriately capture the changes after injection. Because 7 comparisons were made between preinjection and postinjection estimates, a Bonferroni correction was used to control for multiple comparisons and therefore the significance level was adjusted to P ⬍ .007. A similar mixed-effects model was used to model glucose as a function of day and group (IDDM and NIDDM). Because this was an exploratory analysis, no adjustment for multiple comparisons was made.
RESULTS Nineteen subjects initially enrolled in the study. Six patients did not return their blood glucose logs and 1 patient did not complete the blood sugar log as instructed. Therefore, a total of 12 patients (4 men, 8 women, average age 69.5 years) were included in the final statistical analysis (Figure 1). Details of the demographic data are summarized in Table 1. The estimated mean glucose on preinjection days 1-4 was 130 mg/dL, with no change over the preinjection period. The glucose levels peaked on the evening of the injection (236 mg/dL), resulting in a statistically significant elevation of 106 mg/dL (P ⬍ .0001) over the baseline values. The elevation remained statistically significant until 2 days after the injection. The glucose elevation continued throughout the 3-day follow-up period, though these elevations were not statistically significant. Table 2 and Figure 2 demonstrate the estimated mean glucose levels with 95% confidence intervals. During statistical analysis, it was found that NIDDM and IDDM subjects exhibited statistically significant differences in the curves postinjection (P ⫽ .04). The NIDDM group had a larger initial elevation in blood glucose when compared with the IDDM group (P ⫽ .02), though they returned to baseline levels more rapidly (Figure 3). There were no reported complications or adverse effects during the course of this study.
Table 2. Daily glucose levels and differences compared to baseline
Day
Mean Glucose Estimate (mg/dL) (95% CI)
Difference from Days 0-3 (P Value)
⫺3 to 0 0.5 1 1.5 2 2.5 3 3.5
130 (115-145) 236 (207-265) 208 (182-235) 179 (152-206) 171 (142-200) 160 (130-191) 137 (102-173) 166 (128-205)
NA 106 (P ⬍ .0001)* 78 (P ⬍ .0001)* 49 (P ⬍ .0001)* 41 (P ⫽ .0016)* 30 (P ⫽ .03) 7 (P ⫽ .66) 36 (P ⫽ .04)
*Statistically significant after adjusting for multiple comparisons.
PM&R
Vol. 1, Iss. 4, 2009
343
Figure 2. Estimated mean glucose with 95% confidence intervals.
DISCUSSION The current study gives insight into the amount and duration of hyperglycemia seen in diabetic patients after lumbosacral transforaminal and caudal epidural betamethasone injections. There was a statistically significant elevation in blood glucose that peaked on the evening of the injection day at 106 mg/dL above baseline and persisted 2 days after the injection. This increase in blood glucose supports a systemic effect of the corticosteroid, despite epidural administration. The duration of the effect is shorter than has been previously reported after ESIs. However each of these previous studies reported different outcome measures, including postprandial glucose, fasting glucose, and various measures of adrenal axis function [8,9,11,12]. The authors chose home glucometer recordings for the primary outcome measure for this study because glucometers are typically the patient’s most accessible measure of serum blood glucose. The differences found between the NIDDM and IDDM subjects should be viewed with caution given our small study size. There was a large difference between the average age of the NIDDM group (75.3 years) and the IDDM group (63.6 years), and the gender makeup of the groups (1 male and 5 females with NIDDM compared with 3 males and 3 females with
IDDM). Though this heterogeneity of the groups does not allow for conclusive comparisons, the presumption is that the difference may be due to the IDDM subjects using flexible insulin regimens and the NIDDM subjects using only oral medications. Additionally, there were 2 bilateral injections performed among the IDDM patients compared with 1 bilateral injection performed among the NIDDM patients. This may have contributed to the difference seen among the groups. The major limitation in this study is the small sample size (n ⫽ 12), which increases the likelihood of individual statistical outliers overly affecting the final outcomes of the study. Furthermore, this study lacked control over the individual variables surrounding the blood glucose measurements. Specifically, no standardized glucometer was issued to the subjects and the recording times were only described as “AM” or “PM” without regard for the timing of meals and glucometer recordings. The inclusion criteria selected out a population of diabetics with access to a glucometer and a willingness to take their blood sugars twice daily for 7 days. This would suggest a cohort of well-controlled diabetic subjects. Additionally, the study design did not account for the use of as-needed sliding scale or “rescue” insulin. This may have led to lower blood glucose measurements than would have
344
Gonzalez et al
EPIDURAL STEROIDS AND EFFECTS ON BLOOD GLUCOSE
Figure 3. NIDDM vs. IDDM daily blood glucose measurements with mean blood glucose curves. NIDDM, non–insulin-dependent diabetes mellitus; IDDM, insulin-dependent diabetes mellitus.
been otherwise observed in the IDDM subjects. A final limitation of the study was the use of different routes of administration of the betamethasone within the epidural space. It is conceivable that vascular differences between the lumbar spine and the sacrum could have affected the amount of corticosteroid absorbed systemically, thereby altering the amount and duration of hyperglycemia. Given these results, and the results of other recent studies, future designs should have longer follow-up periods and should add outcome measures that more accurately assess the pituitary-adrenal axis [16]. Uniform steroid dosages should be considered in future studies to limit treatment variability. The addition of a nondiabetic patient group and the inclusion of diabetic subjects that do not normally monitor their blood sugars would give better insight into the hyperglycemia that would likely be observed in everyday clinical practice. Further study designs should be advised to use a uniform glucometer with education regarding its use. Specific instructions regarding the glucometer recording times (eg, pre- or postprandial, medication timing) would also limit unnecessary variability in future studies. Additional subject variables should also be addressed, including diet, activity, and the use of medications that blunt or enhance hyperglycemia. This study suggests that lumbosacral TFESIs and caudal ESIs using betamethasone lead to statistically significant elevations in
blood sugar over baseline levels in diabetic subjects. Based on these data, transient hyperglycemia is an expected consequence of these injections and should be discussed with diabetic patients as part of the informed consent process.
REFERENCES 1. Riew KD, Yin Y, Gilula L, Bridwell KH, et al. The effect of nerve-root injections on the need for operative treatment of lumbar radicular pain. A prospective, randomized, controlled, double-blind study. J Bone Joint Surg Am 2000;82A:1589-1593. 2. Weinstein SM, Herring SA. Lumbar epidural steroid injections. Spine J 2003;3(215 Suppl):37S-44S. 3. McLain RF, Kapural L, Mekhail NA. Epidural steroid therapy for back and leg pain: mechanisms of action and efficacy. Spine J 2005;5:191-201. 4. Siddall PJ, Cousins MJ. Spinal pain mechanisms. Spine 1997;22:98104. 5. Flower RJ, Blackwell GJ. Anti-inflammatory steroids induce biosynthesis of a phospholipase A2 inhibitor which prevents prostaglandin generation. Nature 1979;278:456-459. 6. Botwin KP, Gruber RD, Bouchlas CG, Torres-Ramos FM, Freeman TL, Slaten WK. Complications of fluoroscopically guided transforaminal lumbar epidural injections. Arch Phys Med Rehabil 2000;81:1045-1050. 7. Katzung BG. Basic & Clinical Pharmacology. 8th ed. New York: Lange Medical Books/McGraw-Hill; 2001. 8. Ward A, Watson J, Wood P, Dunne C, Kerr D. Glucocorticoid epidural for sciatica: metabolic and endocrine sequelae. Rheumatology (Oxford) 2002;41:68-71. 9. Maillefert JF, Aho S, Huguenin MC, et al. Systemic effects of epidural dexamethasone injections. Rev Rhum Engl Ed 1995;62:429-432.
PM&R
10. Stambough JL, Booth RE, Jr., Rothman RH. Transient hypercorticism after epidural steroid injection. A case report. J Bone Joint Surg Am 1984;66:1115-1116. 11. Kay J, Findling JW, Raff H. Epidural triamcinolone suppresses the pituitary adrenal axis in human subjects. Anesth Analg 1994;79:501-505. 12. Younes M, Neffati F, Touzi M, et al. Systemic effects of epidural and intra-articular glucocorticoid injections in diabetic and nondiabetic patients. Joint Bone Spine 2007;74:472-476. 13. Wang AA, Hutchinson DT. The effect of corticosteroid injection for trigger finger on blood glucose level in diabetic patients. J Hand Surg [Am] 2006;31:979-981.
Vol. 1, Iss. 4, 2009
345
14. Feldman-Billard S, Du Pasquier-Fediaevsky L, Heron E. Hyperglycemia after repeated periocular dexamethasone injections in patients with diabetes. Ophthalmology 2006;113:1720-1723. 15. Habib GS, Abu-Ahmad R. Lack of effect of corticosteroid injection at the shoulder joint on blood glucose levels in diabetic patients. Clin Rheumatol 2007;26:566-568. 16. Linn A, Stoller A, Simopoulos T, Peeters-Asdourian C. Changes in blood glucose levels in diabetics after epidural steroid injection. Abstract presented at: 24th Annual Meeting of the American Academy of Pain Medicine; February 2008; Orlando, FL.
The Effects of Epidural Betamethasone on Blood Glucose in Patients with Diabetes Mellitus Peter Gonzalez, MD, Scott R. Laker, MD, William Sullivan, MD, Jeri E. F. Harwood, PhD, Venu Akuthota, MD Objective: To determine the effects of lumbosacral transforaminal and caudal epidural betamethasone injections on blood glucose levels in diabetic subjects. The hypothesis is that epidural steroid injections result in transient elevation of blood glucose levels in diabetic subjects. Design: This is a prospective, observational cohort. Twelve diabetic subjects (6 non– insulin-dependent and 6 insulin-dependent) receiving lumbosacral or caudal epidural betamethasone injections for neurogenic claudication or radicular pain were studied. Spinal level and approach were decided based on symptoms, pathology, and magnetic resonance imaging findings. Subjects recorded their finger stick blood glucose levels twice daily for 3 days before the injection, the day of the injection, and 3 days after the injection. Setting: A tertiary, university-based, spine center. Participants: Inclusion criteria included diabetic subjects (age 18 years) with the ability and willingness to monitor and report their blood glucose. Exclusion criteria included epidural steroid injections (ESIs) within the previous 2 months or peripheral corticosteroid injections within the previous 2 weeks. Nineteen subjects initially enrolled, and 12 successfully completed the study. Interventions: After informed consent was obtained, subjects underwent fluoroscopically guided lumbosacral transforaminal ESIs (TFESIs) or caudal ESIs, using contrast to confirm targeted needle placement and to rule out vascular uptake. Main Outcome Measures: Subjects recorded morning and evening blood glucose (mg/dL) via glucometer. Results: There was a 106 mg/dL average elevation in blood glucose level on the evening of the injection day. The blood glucose elevation remained statistically significant for 3 days after the injection (P ⬍ .002). Blood glucose remained elevated through postinjection day 2, though these elevations were not statistically significant. Conclusions: Lumbosacral transforaminal and caudal epidural betamethasone injections are associated with statistically significant elevations in blood glucose levels in diabetic subjects. This effect peaked on the day of the injection and lasted approximately 2 days.
INTRODUCTION Epidural steroid injections (ESIs) have been used in the treatment of lumbosacral radicular pain. They have been increasingly used in conjunction with conservative treatment, such as medications and therapeutic exercise to reduce pain in the short term and to obviate surgery [1]. It has been postulated that corticosteroids, such as betamethasone, decrease radicular pain through a reduction of the inflammatory cascade. Proposed mechanisms of action include decreased production of prostaglandins, histamine, leukotrienes, interleukin-1, interleukin-2, interferons, and tumor necrosis factor [2-5]. Hyperglycemia is a reported adverse effect of ESIs [6]. This effect is thought to be due to the opposition of corticosteroid to insulin with increased gluconeogenesis in the liver and decreased glucose uptake in the peripheral tissues [7]. Ward et al [8] reported on insulin sensitivity after ESIs. One previous study reported on glucose tolerance in 10 healthy, nondiabetic individuals receiving caudal ESIs (80 mg triamcinolone) and found that glucose and insulin levels significantly increased 24 hours after injection and returned to normal at PM&R
340
1934-1482/09/$36.00 Printed in U.S.A.
P.G. University of Colorado Denver, School of Medicine, Department of Physical Medicine and Rehabilitation, PO Box 6511, Mail Stop F-493, Aurora, CO 80045. Address correspondence to: P.G.; e-mail: peter.gonzalez@ ucdenver.edu Disclosure: nothing to disclose S.R.L. University of Washington, School of Medicine, Department of Rehabilitation Medicine, Seattle, WA Disclosure: nothing to disclose W.S. University of Colorado Denver, School of Medicine, Department of Physical Medicine and Rehabilitation, Aurora, CO Disclosure: nothing to disclose J.E.F.H. University of Colorado Denver, School of Medicine, Department of Pediatrics, Aurora, CO Disclosure: nothing to disclose V.A. University of Colorado Denver, School of Medicine, Department of Physical Medicine and Rehabilitation, Aurora, CO Disclosure: nothing to disclose Disclosure Key can be found on the Table of Contents and at www.pmrjournal.org Submitted for publication February 20, 2008; accepted December 17, 2008.
© 2009 by the American Academy of Physical Medicine and Rehabilitation Vol. 1, 340-345, April 2009 DOI: 10.1016/j.pmrj.2008.12.007
PM&R
1 week after injection [8]. Maillefert et al [9] reported on the systemic effects of epidural administration of 15 mg of dexamethasone acetate. These authors found that a single injection resulted in transient adrenal suppression indicating passage of corticosteroid into the bloodstream. Though there have been several studies that have reported on endocrinological sequelae of steroid injections, this is one of the first studies that reports directly on blood glucose levels in an exclusively diabetic population receiving lumbosacral transforaminal ESIs (TFESIs) and caudal ESIs. One case reported hypercorticism with Cushingoid features induced by 2 lumbar ESIs in a nondiabetic patient. The patient’s symptoms resolved over the course of 6 weeks [10]. Another study found that the hypothalamic-pituitary axis was altered for an average of 1 month after a series of 3 ESIs given weekly over 3 weeks [11]. The retrospective review by Botwin et al [6] of complications in lumbar ESIs found 1 patient that reported a transient elevation in blood sugar. Most recently, 1 study found that the corticotropic axis was altered up to 21 days after ESIs in both diabetic and nondiabetic patients [12]. Related research on the effects of injectable corticosteroids on hyperglycemia has revealed similar results. A study of patients receiving trigger finger injections with corticosteroid showed a statistically significant hyperglycemic effect on postinjection day 1 and a hyperglycemic trend that lasted 5 days [13]. A study of patients receiving daily periocular steroid injections found a peak in blood glucose 6 hours after the injection and a trough blood glucose level approximately 24 hours later that was 13% higher than baseline levels [14]. Younes et al also found that the corticotropic axis was altered in patient’s receiving intra-articular shoulder injections [12]. However, another study on the effects of intra-articular shoulder steroid injections showed no effect on blood glucose in diabetic patients [15]. The present study was designed to evaluate the effects of epidural betamethasone injections on blood glucose in subjects with diabetes mellitus receiving treatment for lumbar radicular pain. The intent was to quantify the effect on blood glucose and to determine the onset and duration of the changes. The hypothesis was that ESIs result in transient elevation of blood glucose in diabetic subjects.
MATERIALS AND METHODS This study was designed as a prospective, observational cohort that evaluated the effects of lumbosacral TFESIs and caudal ESIs on blood glucose in diabetic patients. The clinical investigators were not blinded given the observational design of the study. Randomization was not performed, because all subjects underwent the same types of interventions. Approval by the Multiple Institutional Review Board was obtained and recruitment of subjects was completed on May 1, 2007. Diabetic patients scheduled to undergo lumbosacral TFESIs or caudal ESIs were invited to participate in the study. All subjects were recruited at a university subspecialty spine
Vol. 1, Iss. 4, 2009
341
Figure 1. Flow diagram.
center setting. Informed consent was obtained by resident, fellow, or attending physicians. Inclusion criteria were subjects with an age greater than 18 years, a medical history of documented insulin-dependent diabetes mellitus (IDDM) or non–insulin-dependent diabetes mellitus (NIDDM), and the ability and willingness to record their blood glucose levels with a glucometer twice daily for 3 days before the injection, the day of the injection, and for 3 days after injection. This follow-up period was determined based on the results of previous studies [8,9,13,14]. Subjects that had received ESIs within the preceding 2 months and those that had received other joint or soft-tissue corticosteroid injections within the preceding 2 weeks were excluded. These time frames were based on previous related research [8-14]. The primary outcome measures were the blood glucose recordings from each subject’s home glucometer. All subjects were provided with a blood glucose log to monitor blood glucose levels each morning
342
Gonzalez et al
EPIDURAL STEROIDS AND EFFECTS ON BLOOD GLUCOSE
Table 1. Demographic data Age
Sex
Type
Diagnosis
Procedure
65 70
M F
IDDM NIDDM
B L5-S1 TFESI S1 TFESI
84
F
NIDDM
55 73 79
M F F
NIDDM IDDM NIDDM
60 77 55 67 87 62
M F F M M F
IDDM NIDDM IDDM IDDM NIDDM IDDM
Stenosis Stenosis with radiculopathy Stenosis with radiculopathy Radiculopathy Stenosis Stenosis with radiculopathy Radiculopathy Radiculopathy Stenosis Stenosis Stenosis Stenosis with radiculopathy
B L4-L5 TFESI L4-L5 TFESI L5-S1 TFESI L4-L5 TFESI L4-L5 TFESI L4-L5 TFESI Caudal ESI B TFESI Caudal ESI L4-L5 TFESI
Abbreviations: M ⫽ male; F ⫽ female; IDDM ⫽ insulin-dependent diabetes mellitus; NIDDM ⫽ non–insulin-dependent diabetes mellitus; TFESI ⫽ transforaminal epidural steroid injection; ESI ⫽ epidural steroid injection; L ⫽ lumbar; B ⫽ bilateral.
and evening. This log was returned to the investigators by mail or at subsequent follow-up appointments. The approach for administration of epidural corticosteroid (lumbosacral transforaminal or caudal approach) and level was decided based on symptoms, spinal pathology, and magnetic resonance imaging findings. All procedures were performed under fluoroscopic guidance by one of three treating investigators (P.G., W.J.S., V.A.). For both lumbosacral TFESIs and caudal ESIs, patients were placed in a prone position on an examination table within the fluoroscopy suite. The skin overlying the target level was cleansed with 10% povidone-iodine, USP prep solution (Novation Inc., Irving, TX). A solution of 1% lidocaine 10 mg/mL (Astra Zeneca; Wilmington, NC) was used to anesthetize the skin and soft tissues. All procedures were performed with 22-gauge, 3.5-inch spinal needles (Sherwood Medical, St. Louis, MO). Needle manipulation was performed under intermittent fluoroscopic guidance. Instillation of iohexol contrast material 240 mgI/mL (GE Healthcare, Oslo, Norway) was performed during each injection to insure proper placement of the needle tip within the epidural space and to confirm no vascular uptake of the injectate. A solution of 1-1.5 mL of 1% lidocaine was administrated in the epidural space prior to the administration of corticosteroid. The corticosteroid used for all injections was 6 mg/mL betamethasone (Schering Corp., Kenilsworth, NJ). Unilateral TFESIs consisted of 2 mL of betamethasone and bilateral injections used 1.5 mL of betamethasone per side, for a total of 3 mL. All caudal ESIs consisted of a mixture of 2 mL of betamethasone, 4 mL of normal saline, and 4 mL of 1% lidocaine.
Statistics All analyses assumed a 2-sided test of hypothesis and were performed in SAS v9.1 (SAS Institute, Inc., Cary, NC). A mixed-effects model was used to model glucose as a function
of day. These models accounted for the correlation between repeated measures on patients. Days ⫺3 through 0 include the 7 glucose measurements obtained preinjection and days 0.5 through 3.5 included all postinjection measurements. The model allowed a slope change at day 0 (ie, at the time of injection) and postinjection glucose was modeled as a function of polynomials of time, to appropriately capture the changes after injection. Because 7 comparisons were made between preinjection and postinjection estimates, a Bonferroni correction was used to control for multiple comparisons and therefore the significance level was adjusted to P ⬍ .007. A similar mixed-effects model was used to model glucose as a function of day and group (IDDM and NIDDM). Because this was an exploratory analysis, no adjustment for multiple comparisons was made.
RESULTS Nineteen subjects initially enrolled in the study. Six patients did not return their blood glucose logs and 1 patient did not complete the blood sugar log as instructed. Therefore, a total of 12 patients (4 men, 8 women, average age 69.5 years) were included in the final statistical analysis (Figure 1). Details of the demographic data are summarized in Table 1. The estimated mean glucose on preinjection days 1-4 was 130 mg/dL, with no change over the preinjection period. The glucose levels peaked on the evening of the injection (236 mg/dL), resulting in a statistically significant elevation of 106 mg/dL (P ⬍ .0001) over the baseline values. The elevation remained statistically significant until 2 days after the injection. The glucose elevation continued throughout the 3-day follow-up period, though these elevations were not statistically significant. Table 2 and Figure 2 demonstrate the estimated mean glucose levels with 95% confidence intervals. During statistical analysis, it was found that NIDDM and IDDM subjects exhibited statistically significant differences in the curves postinjection (P ⫽ .04). The NIDDM group had a larger initial elevation in blood glucose when compared with the IDDM group (P ⫽ .02), though they returned to baseline levels more rapidly (Figure 3). There were no reported complications or adverse effects during the course of this study.
Table 2. Daily glucose levels and differences compared to baseline
Day
Mean Glucose Estimate (mg/dL) (95% CI)
Difference from Days 0-3 (P Value)
⫺3 to 0 0.5 1 1.5 2 2.5 3 3.5
130 (115-145) 236 (207-265) 208 (182-235) 179 (152-206) 171 (142-200) 160 (130-191) 137 (102-173) 166 (128-205)
NA 106 (P ⬍ .0001)* 78 (P ⬍ .0001)* 49 (P ⬍ .0001)* 41 (P ⫽ .0016)* 30 (P ⫽ .03) 7 (P ⫽ .66) 36 (P ⫽ .04)
*Statistically significant after adjusting for multiple comparisons.
PM&R
Vol. 1, Iss. 4, 2009
343
Figure 2. Estimated mean glucose with 95% confidence intervals.
DISCUSSION The current study gives insight into the amount and duration of hyperglycemia seen in diabetic patients after lumbosacral transforaminal and caudal epidural betamethasone injections. There was a statistically significant elevation in blood glucose that peaked on the evening of the injection day at 106 mg/dL above baseline and persisted 2 days after the injection. This increase in blood glucose supports a systemic effect of the corticosteroid, despite epidural administration. The duration of the effect is shorter than has been previously reported after ESIs. However each of these previous studies reported different outcome measures, including postprandial glucose, fasting glucose, and various measures of adrenal axis function [8,9,11,12]. The authors chose home glucometer recordings for the primary outcome measure for this study because glucometers are typically the patient’s most accessible measure of serum blood glucose. The differences found between the NIDDM and IDDM subjects should be viewed with caution given our small study size. There was a large difference between the average age of the NIDDM group (75.3 years) and the IDDM group (63.6 years), and the gender makeup of the groups (1 male and 5 females with NIDDM compared with 3 males and 3 females with
IDDM). Though this heterogeneity of the groups does not allow for conclusive comparisons, the presumption is that the difference may be due to the IDDM subjects using flexible insulin regimens and the NIDDM subjects using only oral medications. Additionally, there were 2 bilateral injections performed among the IDDM patients compared with 1 bilateral injection performed among the NIDDM patients. This may have contributed to the difference seen among the groups. The major limitation in this study is the small sample size (n ⫽ 12), which increases the likelihood of individual statistical outliers overly affecting the final outcomes of the study. Furthermore, this study lacked control over the individual variables surrounding the blood glucose measurements. Specifically, no standardized glucometer was issued to the subjects and the recording times were only described as “AM” or “PM” without regard for the timing of meals and glucometer recordings. The inclusion criteria selected out a population of diabetics with access to a glucometer and a willingness to take their blood sugars twice daily for 7 days. This would suggest a cohort of well-controlled diabetic subjects. Additionally, the study design did not account for the use of as-needed sliding scale or “rescue” insulin. This may have led to lower blood glucose measurements than would have
344
Gonzalez et al
EPIDURAL STEROIDS AND EFFECTS ON BLOOD GLUCOSE
Figure 3. NIDDM vs. IDDM daily blood glucose measurements with mean blood glucose curves. NIDDM, non–insulin-dependent diabetes mellitus; IDDM, insulin-dependent diabetes mellitus.
been otherwise observed in the IDDM subjects. A final limitation of the study was the use of different routes of administration of the betamethasone within the epidural space. It is conceivable that vascular differences between the lumbar spine and the sacrum could have affected the amount of corticosteroid absorbed systemically, thereby altering the amount and duration of hyperglycemia. Given these results, and the results of other recent studies, future designs should have longer follow-up periods and should add outcome measures that more accurately assess the pituitary-adrenal axis [16]. Uniform steroid dosages should be considered in future studies to limit treatment variability. The addition of a nondiabetic patient group and the inclusion of diabetic subjects that do not normally monitor their blood sugars would give better insight into the hyperglycemia that would likely be observed in everyday clinical practice. Further study designs should be advised to use a uniform glucometer with education regarding its use. Specific instructions regarding the glucometer recording times (eg, pre- or postprandial, medication timing) would also limit unnecessary variability in future studies. Additional subject variables should also be addressed, including diet, activity, and the use of medications that blunt or enhance hyperglycemia. This study suggests that lumbosacral TFESIs and caudal ESIs using betamethasone lead to statistically significant elevations in
blood sugar over baseline levels in diabetic subjects. Based on these data, transient hyperglycemia is an expected consequence of these injections and should be discussed with diabetic patients as part of the informed consent process.
REFERENCES 1. Riew KD, Yin Y, Gilula L, Bridwell KH, et al. The effect of nerve-root injections on the need for operative treatment of lumbar radicular pain. A prospective, randomized, controlled, double-blind study. J Bone Joint Surg Am 2000;82A:1589-1593. 2. Weinstein SM, Herring SA. Lumbar epidural steroid injections. Spine J 2003;3(215 Suppl):37S-44S. 3. McLain RF, Kapural L, Mekhail NA. Epidural steroid therapy for back and leg pain: mechanisms of action and efficacy. Spine J 2005;5:191-201. 4. Siddall PJ, Cousins MJ. Spinal pain mechanisms. Spine 1997;22:98104. 5. Flower RJ, Blackwell GJ. Anti-inflammatory steroids induce biosynthesis of a phospholipase A2 inhibitor which prevents prostaglandin generation. Nature 1979;278:456-459. 6. Botwin KP, Gruber RD, Bouchlas CG, Torres-Ramos FM, Freeman TL, Slaten WK. Complications of fluoroscopically guided transforaminal lumbar epidural injections. Arch Phys Med Rehabil 2000;81:1045-1050. 7. Katzung BG. Basic & Clinical Pharmacology. 8th ed. New York: Lange Medical Books/McGraw-Hill; 2001. 8. Ward A, Watson J, Wood P, Dunne C, Kerr D. Glucocorticoid epidural for sciatica: metabolic and endocrine sequelae. Rheumatology (Oxford) 2002;41:68-71. 9. Maillefert JF, Aho S, Huguenin MC, et al. Systemic effects of epidural dexamethasone injections. Rev Rhum Engl Ed 1995;62:429-432.
PM&R
10. Stambough JL, Booth RE, Jr., Rothman RH. Transient hypercorticism after epidural steroid injection. A case report. J Bone Joint Surg Am 1984;66:1115-1116. 11. Kay J, Findling JW, Raff H. Epidural triamcinolone suppresses the pituitary adrenal axis in human subjects. Anesth Analg 1994;79:501-505. 12. Younes M, Neffati F, Touzi M, et al. Systemic effects of epidural and intra-articular glucocorticoid injections in diabetic and nondiabetic patients. Joint Bone Spine 2007;74:472-476. 13. Wang AA, Hutchinson DT. The effect of corticosteroid injection for trigger finger on blood glucose level in diabetic patients. J Hand Surg [Am] 2006;31:979-981.
Vol. 1, Iss. 4, 2009
345
14. Feldman-Billard S, Du Pasquier-Fediaevsky L, Heron E. Hyperglycemia after repeated periocular dexamethasone injections in patients with diabetes. Ophthalmology 2006;113:1720-1723. 15. Habib GS, Abu-Ahmad R. Lack of effect of corticosteroid injection at the shoulder joint on blood glucose levels in diabetic patients. Clin Rheumatol 2007;26:566-568. 16. Linn A, Stoller A, Simopoulos T, Peeters-Asdourian C. Changes in blood glucose levels in diabetics after epidural steroid injection. Abstract presented at: 24th Annual Meeting of the American Academy of Pain Medicine; February 2008; Orlando, FL.