Outcomes following first-time lower extremity revascularization for chronic limb-threatening ischemia in insulin-dependent diabetic patients

Abstract

OBJECTIVES

Historically, open surgical bypass provided a durable repair among diabetic patients with chronic limb-threatening ischemia (CLTI). In the current endovascular era, however, the difference in long-term outcomes between first-time revascularization strategies among patients with insulin-dependent diabetes (IDDM) is poorly understood.

METHODS

We reviewed the records of all patients with IDDM undergoing a first-time infrainguinal bypass (BPG) or percutaneous transluminal angioplasty (PTA/S) for CLTI at our institution from 2005–2014. We defined IDDM as use of chronic insulin administration at baseline to control blood glucose levels and recorded the most recent HbA1c value available within 3 months pre-procedure and fasting blood glucose (FBG) levels on the day of procedure. We compared rates of wound healing, restenosis, re-intervention, major amputation, and mortality between BPG and PTA/S in our population using Chi-square, Kaplan-Meier, and Cox regression analyses. As a sensitivity analysis, we calculated propensity scores and employed inverse probability weighting to account for nonrandom assignment to BPG versus PTA/S.

RESULTS

Of 2,869 infrainguinal revascularizations from 2005–2014, 655 limbs (316 BPG, 339 PTA/S) in 580 patients fit our criteria and underwent a first-time revascularization for CLTI. Patients undergoing bypass, as compared to PTA/S, were similar in age (69 vs. 68 years; P=.55) and had similar rates of tissue loss (87% vs. 91%; P=.07) and dialysis-dependence (26% vs. 28%; P=.55), were less likely to be hypertensive (84% vs. 92%; P<.001), and were more likely to be current smokers (21% vs. 14%; P=.02). There were no differences between BPG and PTA/S patients related to mean HbA1c levels (8.1 vs. 8.0; P=.51) or mean FBG levels (158 vs. 150; P=.18). Although total hospital length of stay was significantly longer among BPG patients (11 vs. 8 days; P<.001), perioperative complications did not differ, including acute kidney injury (19% vs. 23%; P=.24), hematoma (6.0% vs. 3.8%; P=.20), acute myocardial infarction (1.3% vs. 2.1%; P=.43), and mortality (3.8% vs. 3.0%; P=.55). BPG-first patients had significantly lower unadjusted 6-month rates of incomplete wound healing (49% vs. 57%) and 5-year rates of restenosis (53% vs. 72%) and re-intervention (47% vs. 58%) (all P<.05). After adjustment, multivariable analysis suggested PTA/S-first intervention to be significantly associated with higher risk of restenosis (Hazard Ratio (HR), 1.9; 95% Confidence Interval [1.3–2.7]) and re-intervention (1.9 [1.2–2.7]). These results remained robust following inverse probability weighting.

CONCLUSION

Among patients with insulin-dependent diabetes and CLTI, a bypass first strategy is associated with similar 30-day outcomes and lower restenosis and re-intervention rates. These data suggest that a bypass-first approach may best serve appropriately selected, anatomically suitable patients with IDDM and pedal ischemia that requires revascularization.

INTRODUCTION

Peripheral arterial disease (PAD) is an independent risk factor for cardiovascular morbidity and mortality, and affects up to 10% of adults around the world – a number that continues to rise.^1–3 Concurrently, the prevalence and incidence of diabetes – a condition that significantly increases the likelihood of developing PAD – effectively doubled between 1990–2008.^{3, 4} Chronic limb-threatening ischemia (CLTI), the most advanced stage of PAD and a debilitating condition that severely impacts quality of life and long-term survival, occurs more frequently in diabetic patients.^2–4 While proper diabetes management is crucial in preventing the occurrence and progress of PAD, optimal revascularization strategies, when necessary, play an equally integral role in changing the course of the disease.

Conflicting evidence continues to amplify controversy regarding which treatment, open surgical bypass (BPG) or percutaneous transluminal angioplasty with or without stenting (PTA/S), is associated with a better clinical outcome, especially in particularly prevalent and vulnerable populations. Our previous study suggested that insulin-dependent diabetes (IDDM) in patients undergoing a first-time revascularization for CLTI conferred an elevated risk for a number of adverse long-term outcomes, including longer hospital lengths of stay, incomplete wound healing, major amputation, and major adverse limb events.⁵ Although an endovascular-first strategy is becoming more common, these prior findings prompted us to ask whether this strategy remained beneficial for patients with IDDM and CLTI. We thus evaluated the perioperative and long-term differences between BPG and PTA/S outcomes among patients with IDDM undergoing a first-time revascularization for CLTI in our institution.

METHODS

Subjects and settings

We performed a retrospective review of all patients with CLTI undergoing a first-time infrainguinal intervention at Beth Israel Deaconess Medical Center (BIDMC). Medical records of all BPG and PTA/S interventions from January 2005 to October 2014 were individually reviewed. We defined insulin-dependent diabetes (IDDM) as preoperative or at-home insulin administration to control blood glucose levels. The most recent HbA1c value available within 3 months pre-procedure and fasting blood glucose (FBG) levels on the day of procedure were recorded. We excluded patients without diabetes and those with diabetes who were not prescribed insulin. We also excluded patients who received previous interventions on the ipsilateral limb or interventions proximal to the common femoral artery (whether at BIDMC or at an outside institution). Importantly, insulin dependence, for the purposes of this study, is not considered equivalent to type I diabetes, as it describes the patient-level pattern of insulin use at the time of revascularization. We included patients undergoing a concomitant procedure, including endarterectomy, profundaplasty, thrombectomy, or patch, and adjusted for these in our multivariable analyses. Multiple-vessel PTA/S interventions (i.e., more than one vessel within one procedure) were performed throughout our study period; as such, PTA/S interventions on individual vessels should not be considered as mutually exclusive events. Typical patient follow-up consisted of visits every 3–4 months for two years and every six months afterwards, with arterial duplex ultrasound imaging and ankle-brachial indices with forefoot pulse volume recordings and/or toe pressures.

Our analysis included patients whose indication for intervention was distinctly classifiable as CLTI and who underwent either a first-time BPG or a first-time PTA/S. Indications for intervention included tissue loss (i.e., gangrene and ulcer) or rest pain. Femoropopliteal lesion anatomy and severity were defined according to the modified Trans Atlantic Inter-society Consensus (TASC II) classification. As the modified TASC II classification does not include an updated TASC class for tibial lesions, tibial lesion information was defined by TASC I.^{1, 6} Two surgeons who were not investigators on this study retrospectively reviewed all angiograms for lesion severity and TASC classification.

Measurements and outcome variables

Primary outcomes included perioperative complications, wound healing, restenosis, re-intervention, major amputation, and mortality. Secondary outcomes included RAS events (a composite variable denoted by re-intervention, major amputation, or stenosis) and major adverse limb events (MALE, a composite variable denoted by any major amputation or creation of a new bypass graft, a jump graft revision, surgical thrombectomy with or without surgical patch angioplasty, and thrombectomy of an occluded graft or arterial segment using pharmacologic or mechanical thrombolysis).⁷ Demographics, comorbidities, and SVS WIfI information were retrospectively assessed based on operative reports, discharge summaries, imaging, and progress reports available pre-operatively and throughout the follow-up period.⁸ Comorbidities were defined according to the JVS reporting standards. Perioperative complications included hematoma, acute kidney injury, myocardial infarction, stroke, and death. We defined restenosis as >70% by angiography measurement or >3.5-fold increase in peak systolic velocity by duplex. Re-intervention, which includes any ipsilateral surgical or endovascular revision, was performed for symptomatic graft restenosis and threatened asymptomatic grafts. In general, patients did not undergo re-intervention for an asymptomatic restenosis post PTA alone; however, re-intervention was more likely with PTA/S for an asymptomatic in-stent restenosis with peak systolic velocity ratio of >3.5–4. Following BPG, patients were prescribed aspirin and a statin, and were not routinely prescribed Plavix. Anticoagulation and cilostazol use was attending-dependent and varied with operative findings. Additionally, patients undergoing an infrainguinal PTA/S received dual antiplatelet therapy for 30 days followed by aspirin indefinitely. Routine statin use was introduced over time.⁹ Technical success following PTA/S was defined as less than 30% residual stenosis and no flow-limiting dissection, and technical success following BPG included a patent bypass graft at completion, defined as one without significant defect in the vein on angioscopy and continuous wave Doppler interrogation. BPG patients routinely received preoperative vein mapping, angioscopy, and general anesthesia.¹⁰ Complete wound healing was determined by the attending surgeon at follow-up. All mortality information was confirmed through the Social Security Death Index.¹¹

Statistical Analyses

For limb events (i.e., wound healing, stenosis, re-intervention, amputation), we analyzed limbs independently; however, for mortality we censored the initial limb at the procedure date of the contralateral limb to avoid repeated measures. We compared categorical variables using Pearson chi-square and Fisher exact tests and continuous variables using Student t-test or Mann-Whitney U test where appropriate. We constructed Kaplan-Meier curves for long-term outcomes, and compared groups using the log-rank test. For our multivariable analysis, we utilized Cox regression. Covariates were selected using purposeful selection, including both backward selection after a univariate screen (P < .10) and relevant patient factors previously identified.¹² Multivariable Cox regression models were constructed to assess independent associations between procedure type and time-dependent outcomes. Statistical significance was defined as P < .05.

As a sensitivity analysis, we calculated propensity scores for all patients and employed inverse probability weighting to account for nonrandom assignment to procedure type (i.e., BPG and PTA/S). Covariates were generously introduced into the model and included index procedure (BPG vs. PTA/S), procedure year, age, sex, race, hypertension, renal disease, coronary artery disease, diabetes, chronic obstructive pulmonary disease, current smoking, heart failure, indication for intervention, and preoperative ambulatory status. The propensity scores were used to create inverse probability weights (the inverse of the probability of receiving the surgical option that the subject received). These scores were tested for adequacy of overlap by plotting the propensity score distributions between BPG and PTA/S arms. After weighting, the standardized differences were all less than 10% (the usual threshold), indicating minimal imbalance.

All statistical tests utilized STATA 13 and 14.1 (StataCorp, College Station, Tex). The Beth Israel Deaconess Medical Center Institutional Review Board approved this study and waived the need for patient consent due to the retrospective design.

RESULTS

Baseline Characteristics

Of the 2,869 total lower extremity revascularizations performed between January 2005 and October 2014, 667 were on patients with non-CLTI symptoms, 475 were re-interventions, and 433 were on patients who underwent interventions prior to 2005 and were thus excluded from this analysis. Among the remaining 1,294 first-time lower extremity procedures for CLTI, 655 (51%) fit our criteria for IDDM (377 noninsulin-dependent and 262 non-diabetic patients were excluded). Among all IDDM patients within this analysis, 22% had Type I diabetes (N = 145), 77% had Type II diabetes (N = 503), and 1% had post-transplant diabetes (N = 7) (Table I). As Figure 1 illustrates, the number of IDDM patients treated with a revascularization gradually decreased over the study period (from 84 procedures in 2005 to 58 in 2013), as did the proportion of IDDM patients treated with a BPG-first approach (from 79% in 2005 to 22% in 2013).

Table I:

Demographics and co-morbidities in 655 insulin-dependent diabetic patients undergoing either a first-time bypass (BPG) or percutaneous transluminal angioplasty (PTA/S) for chronic limb-threatening ischemia (CLTI) between 2005–2014

	BPG (N=316)	PTA/S (N=339)	P-value
Demographics No. (%)
Age, mean (SD)	68.9 (12.3)	68.3 (11.8)	.55
Nonwhite race	67 (21)	106 (32)	<.01
Male sex	202 (64)	208 (61)	.50
Indication, No. (%)
Rest Pain	42 (13)	30 (8.9)	.07
Ulcer	167 (53)	209 (62)	.03
Gangrene	107 (34)	100 (30)	.23
Tissue loss	274 (87)	309 (91)	.07
Comorbidities, No. (%)
Diabetes
Type I	64 (20)	68 (21)	.81
Type II	243 (78)	260 (78)	.98
Post-transplant	5 (1.6)	2 (0.6)	.27
Coronary artery disease	185 (59)	184 (55)	.27
Hypertension	265 (84)	312 (92)	.001
HbA1c, mean	8.1	8	.51
Fasting blood glucose, mean	158	150	.18
Hyperlipidemia	202 (64)	235 (69)	.14
Dialysis dependence	82 (26)	95 (28)	.55
History of myocardial infarction	106 (34)	100 (30)	.24
Congestive heart failure	116 (37)	105 (31)	.11
COPD	37 (12)	28 (8.3)	.14
Current smoker	67 (21)	48 (14)	.02
Smoking history	197 (63)	177 (52)	<.01
WIfI clinical stage, No. (%)
Clinical stage 1	2 (0.9)	3 (1.1)	.84
Clinical stage 2	46 (21)	47 (18)	.29
Clinical stage 3	55 (26)	80 (30)	.30
Clinical stage 4	112 (52)	138 (52)	.90
Wound position, No. (%)
Forefoot	152 (61)	176 (62)	.72
Midfoot	35 (14)	30 (11)	.29
Heel	52 (19)	59 (19)	.97
Ankle, shin, calf	10 (3.7)	19 (6.2)	.17
Femoropopliteal TASC classification, No. (%)
TASC A	51 (23)	60 (27)	.94
TASC B	69 (30)	106 (47)	.049
TASC C	48 (21)	25 (11)	<.001
TASC D	59 (26)	35 (16)	<.001
Tibial TASC classification, No. (%)
TASC A	24 (11)	47 (16)	.053
TASC B	53 (24)	82 (27)	.14
TASC C	45 (20)	83 (28)	.01
TASC D	101 (45)	89 (30)	<.01

BPG, bypass grafting; PTA/S, percutaneous transluminal angioplasty with or without stenting; SD, standard deviation; HbA1c, glycated hemoglobin; COPD, chronic obstructive pulmonary disease; WIfI, wound, ischemia, foot infection; TASC, Trans-Atlantic Inter-Society Consensus

Figure 1:

Number of yearly first-time revascularization procedures performed on insulin-dependent diabetic patients (IDDM) with chronic limb-threatening ischemia

BPG, open surgical bypass; PTA/S, percutaneous angioplasty with or without stent; IDDM, insulin-dependent diabetes

Overall, BPG and PTA/S patients were similar in age (69 vs. 68 years; P = .55) and had similar rates of tissue loss (87% vs. 91%; P = .07), coronary artery disease (59% vs. 55%; P = .27), and dialysis-dependence (26% vs. 28%; P = .55). Conversely, BPG-first patients were less likely to have hypertension (84% vs. 92%; P = .001) and more likely to have used tobacco products at any point (63% vs. 52%; P < .01) and at the time of revascularization (21% vs. 14%; P = .02). No differences existed between BPG and PTA/S in regards to mean HbA1c levels (8.1 vs. 8.0; P = .51) or mean FBG levels (158 vs. 150; P = .18). Preoperative femoropopliteal TASC D lesions were more common in BPG-first patients (26% vs. 16%; P < .001), as were tibial TASC D lesions (45% vs. 30%; P < .01). We found no difference in the proportion of BPG or PTA/S patients presenting with WIfI clinical stage 4 (52% vs. 52%; P = .90), which was noted to be the most common stage for each group. Lastly, among patients with tissue loss (N=583; 89%), there were no differences between BPG-first and PTA/S-first limbs in regards to anatomic position of wounds, including forefoot (61% vs. 62%; P = .72), midfoot (14% vs. 11%; P = .29), heel (19% vs. 19%; P = .97), or ankle, shin, and calf wounds (3.7% vs. 6.2%; P = .17).

Of the 316 BPG-first procedures, the femoral artery was the most common inflow artery (68% of all procedures), with the dorsalis pedis artery being the most common outflow vessel (31%) (Table II). We used single-segment great saphenous vein conduits in over three-quarters of procedures performed (82%), where non-reversed great saphenous vein was the most common conduit (42%). We utilized synthetic conduits in 10% of procedures and composite vein in approximately 6%. Concomitant procedures included endarterectomy (N=29), patch (N=21) profundoplasty (N=2), and thrombectomy (N=2).

Table II:

Operative characteristics of 316 insulin-dependent diabetic patients undergoing open surgical bypass (BPG) for chronic limb-threatening ischemia (CLTI) between 2005–2014

	No.	%
Inflow artery
Femoral	214	68
Popliteal	102	32
Outflow artery
Above-knee popliteal	46	15
Below-knee popliteal	39	12
Anterior tibial	46	15
Posterior tibial	70	22
Peroneal	17	5.4
Dorsalis pedis/pedal	98	31
Inflow-Outflow
Femoral-above-knee popliteal	45	14
Femoral-below-knee popliteal	38	12
Femoral-anterior tibial	36	11
Femoral-posterior tibial	42	13
Femoral-peroneal	9	2.8
Femoral-dorsalis pedis/pedal	44	14
Popliteal-anterior tibial	10	3.2
Popliteal-posterior tibial	28	8.9
Popliteal-peroneal	7	2.2
Popliteal-dorsalis pedis/pedal	57	18
Conduit
In situ saphenous vein	72	23
Reversed saphenous vein	38	12
Non-reversed saphenous vein	134	42
Arm vein	22	7.0
Composite vein	18	5.7
Synthetic	32	10

Finally, of the 339 PTA/S-first procedures, the superficial femoral artery was the most common proximal vessel intervened on (57%), while we performed 30% of endovascular procedures on the anterior tibial artery (Table III). Overall, 31% of PTA/S procedures were multi-level and 45% of procedures included stenting (53% in femoropopliteal vessels and 13% in infrapopliteal vessels). Concomitant procedures included endarterectomy (N=4), patch (N=4), and thrombectomy (N=1). Within this cohort, there was no use of atherectomies, drug-coated balloons, or drug-eluting stents.

Table III:

Operative characteristics of 339 insulin-dependent diabetic patients undergoing percutaneous transluminal angioplasty with or without stenting (PTA/S) for chronic limb-threatening ischemia (CLTI) between 2005–2014

	No.	%
Proximal vessel
Femoral	193	57
Popliteal	113	33
Femoropopliteal only	126	37
Infrapopliteal vessel
Anterior tibial	103	30
Posterior tibial	54	16
Peroneal	69	20
Dorsalis pedis/pedal	8	2.4
Tibial only	107	32
Multi-level (fem-pop + infra-pop)	106	31
Stenting
Any	151	45
Femoropopliteal	123	53
Infrapopliteal	28	13

The median follow-up for BPG and PTA/S patients was 1.8 years (range <1–10) and 1.5 years (<1–8), respectively.

Perioperative Outcomes

Following first-time lower extremity revascularization for CLTI in patients with IDDM, BPG-first and PTA/S-first patients experienced similar rates of perioperative mortality (3.8% vs. 3.0%; P = .55) (Table IV). Perioperative complications, including stroke (0.6% vs. 0.6%; P = .94), acute myocardial infarction (1.3% vs. 2.1%; P = .43), hematoma (6.0% vs. 3.8%; P = .20), and acute kidney injury (19% vs. 23%; P = .24), did not differ between groups. Perioperative surgical site infections occurred in 11% of BPG-first interventions, and BPG-first patients experienced significantly longer total hospital length of stay (11 vs. 8 days; P < .001). After adjusting for baseline characteristics, multivariable analysis showed that procedure type was not associated with perioperative complications or death (Supplementary Table I).

Table IV:

Perioperative outcomes and complications in 655 insulin-dependent diabetic patients undergoing either a first-time bypass (BPG) or percutaneous transluminal angioplasty (PTA/S) for chronic limb-threatening ischemia (CLTI) between 2005–2014

	BPG (N=316)	PTA/S (N=339)
Perioperative complications, No. (%)
Mortality	12 (3.8)	10 (3.0)	.55
Acute myocardial infarction	4 (1.3)	7 (2.1)	.43
Stroke	2 (0.6)	2 (0.6)	.94
Acute kidney injury	60 (19)	77 (23)	.24
Hematoma	19 (6.0)	13 (3.8)	.20
Surgical site infection	36 (11)	N/A	N/A
Length of stay, mean days
Pre-operative	3.8	3.0	.01
Post-operative	7.3	5.3	<.001
Total	10.9	8.2	<.001

BPG, open surgical bypass; PTA/S, percutaneous transluminal angioplasty with or without stenting;

Long-term Outcomes

Unadjusted Kaplan-Meier analysis demonstrated that complete wound healing at 6-month follow-up was significantly better following BPG-first interventions (51% vs. 43%; P = .01) (Table V). Unadjusted Kaplan-Meier analyses further illustrated BPG-first interventions to have significantly higher 5-year incidences of freedom from restenosis (47% vs. 28%; P < .001) (Figure 2), re-intervention (53% vs. 42%; P = .04) (Figure 3), and RAS events (33% vs. 25%; P < .001) (Figure 4), but did not demonstrate any differences in 5-year rates of limb salvage (70% vs. 74%; P = .94), freedom from MALE (55% vs. 69%; P = .29), or survival (36% vs. 29%; P = .23). Among PTA/S patients experiencing a restenosis throughout follow-up (N = 141), there was no difference between femoropopliteal (N = 75) and tibial restenosis (N = 66) (P = .69). Further, in regards to specific re-intervention type (e.g., open versus endovascular), between BPG and PTA/S patients, Kaplan-Meier analysis illustrated that open re-interventions at 5 years occurred in 24% and 18%, respectively (logrank P = .97), while endovascular re-interventions at 5 years occurred in 31% vs. 50%, respectively (P = .02). Mean number of re-interventions within 1 year of the index procedure did not differ between BPG and PTA/S groups (.47 vs. .59; P = .17), but was trending towards a significant difference 5 years after the index procedure (.39 vs. .51; P = .054). Lastly, throughout follow-up, 7% of BPG-first patients underwent 2 or more re-interventions, while 10% of PTA/S-first patients underwent 2 or more (P = .46).

Table V:

Unadjusted and adjusted long-term outcomes in insulin-dependent diabetic patients undergoing either open surgical bypass (BPG) or percutaneous transluminal angioplasty/stenting (PTA/S) for chronic limb-threatening ischemia (CLTI)

Five-year outcomes	BPG (N=316)	PTA/S (N=339)	Logrank P-value	Adjusted P-value*
	%	%
Wound healing (6 mo.)	51	43	.02	.33
Restenosis	53	72	<.001	<.01
Re-intervention	47	58	.04	<.01
Open	24	18	.97	.10
Endovascular	31	50	.02	.03
Major amputation	30	26	.94	.90
RAS events	67	75	.001	<.001
Major adverse limb events	45	31	.29	.21
Mortality	64	71	.23	.20

BPG, open surgical bypass; PTA/S, percutaneous transluminal angioplasty; RAS, re-intervention, amputation, stenosis

^*Adjusted for procedure year, race, sex, age, coronary artery disease, hypertension, dialysis dependence, congestive heart failure, chronic obstructive pulmonary disease, stenting, history of smoking, concomitant procedures, ambulatory status, indication for intervention, WIfI mean score, femoropopliteal TASC class, tibial TASC class

Figure 2:

Unadjusted freedom from restenosis in insulin-dependent diabetic patients undergoing either open surgical bypass (BPG) or percutaneous transluminal angioplasty/stenting (PTA/S) for chronic limb-threatening ischemia (CLTI)

S.E., standard error

Figure 3:

Unadjusted freedom from re-intervention in insulin-dependent diabetic patients undergoing either open surgical bypass (BPG) or percutaneous transluminal angioplasty/stenting (PTA/S) for chronic limb-threatening ischemia (CLTI)

S.E., standard error

Figure 4:

Unadjusted freedom from restenosis, re-intervention, or amputation (RAS) events in insulin-dependent diabetic patients undergoing either open surgical bypass (BPG) or percutaneous transluminal angioplasty/stenting (PTA/S) for chronic limb-threatening ischemia (CLTI)

S.E., standard error

After adjustment, among all procedure types, PTA/S-first interventions in IDDM patients were shown to be independently associated with restenosis (Hazard Ratio (HR) 1.9, 95% Confidence Interval [CI], 1.3–2.7), any re-intervention (1.8 [1.2–2.7]), endovascular-specific re-intervention (1.8 [1.1–3.0]), and RAS events (1.8 [1.3–2.5]). Incomplete wound healing was no longer associated with procedure type (1.2 [0.8–1.7]), but it did remain associated with hypertension (1.8 [1.1–3.0]). Notably, independent of procedure type, IDDM patients with tibial TASC D lesions (compared to those with TASC A lesions) experienced a higher risk of undergoing a re-intervention (2.1 [1.02–4.3]), RAS event (2.7 [1.6–4.6]), major amputation (3.0 [1.02–8.9]), and MALE (3.7 [1.5–9.0]). Wheelchair-dependence was associated with major amputation (3.6 [1.6–7.9]), while male sex and congestive heart failure independently heightened a patient’s risk of mortality (1.7 [1.1–2.6]) and 1.9 [1.3–2.7], respectively). Neither HbA1c nor fasting blood glucose levels were associated with any long-term outcome of interest.

Sensitivity Analysis

As a means to account for nonrandom assignment to intervention type, we constructed propensity scores for assignment to BPG versus PTA/S and applied inverse probability weighting. In this sensitivity analysis, PTA/S-first interventions remained associated with restenosis (P < .001), re-intervention (P < .01), and RAS events (P < .01). The results of our primary analysis remained robust, as this sensitivity analysis did not appreciably alter any short- or long-term outcomes.

DISCUSSION

Our data suggest that open surgical bypass provides a more durable revascularization strategy for IDDM patients undergoing a first-time lower extremity procedure for CLTI. However, BPG is associated with high rates of wound infection and longer hospital length of stay. Although there are no differences between revascularization strategies related to short- or long-term patient-based risk (i.e., mortality), BPG is associated with statistically lower rates of limb-based risk, including restenosis and re-intervention through follow-up. Importantly, our data illustrate that IDDM patients typically present with severe disease despite their relatively young age, with over three-quarters of patients within our analysis presenting with tissue loss and hypertension, well over half with a history of smoking, hyperlipidemia, and coronary artery disease, and over one-quarter with a history of myocardial infarction and end-stage renal disease. Ultimately, this study builds on our prior work and suggests that the oft-referenced adverse outcomes in IDDM patients may be most mitigated following a BPG-first strategy.¹³

Prior studies similarly demonstrated a trend towards more severe disease in the CLTI population over time, most notably Agarwal et al. in 2016.¹⁴ Using the Nationwide Inpatient Sample (NIS), they characterized US hospital trends in over 640,000 US patients with CLTI between 2003–2011 and discovered that the number of CLTI patients with diabetes, chronic kidney disease, obesity, and prior amputations monotonously increased annually since 2003. During this same period, however, the number of surgical revascularizations steadily decreased. Originally, many saw the endovascular approach as an effective means to treat those with poor surgical alternatives, which has been supported over the years; however, data from several recent studies have suggested that a PTA/S-first strategy may ultimately prove less durable among select patients with CLTI that were potential candidates for either procedure.^{13, 15–18} To date, few studies addressed these concerns in IDDM patients specifically – one of the more diseased and prevalent vascular populations.^{19, 20} Ultimately, given the ability to focus on this important subset of patients, our study suggests that use of a BPG-first strategy, when appropriate, may provide a more durable repair.

Although within a distinct population, this study contributes to the ever-growing literature regarding the preferred primary revascularization strategy for CLTI. Previous studies attempted to compare bypass and PTA/S for various degrees of lower extremity limb ischemia with varying methodological problems.^31–41 Most saliently, the randomized BASIL trial concluded that bypass-first and angioplasty-first strategies provided equal outcomes for amputation-free survival up to two years, after which overall survival and amputation-free survival were better following bypass.¹⁵ Although the BASIL trial has been criticized for its strict eligibility requirements and low number of patients with infrapopliteal disease, our prior work regarding long-term differences between 668 BPG-first and 668 PTA/S-first interventions for CLTI addressed some of these limitations and supported the conclusions that a BPG-first strategy provides a more durable repair in patients with CLTI, especially in those who are fit and have a suitable single-segment saphenous vein conduit.⁴² The current study extends this recommendation to IDDM patients with CLTI, where those with both conditions who are fit and suitable for a bypass-first strategy should expect a more durable repair than those undergoing a PTA/S-first intervention.

As our findings describe, IDDM patients undergoing a first-time PTA/S procedure for CLTI were associated with increased re-intervention and restenosis, but not increased MALE, illustrating that the majority of these re-interventions were minor in nature. This finding is an expected one, as minor re-interventions, especially following a failed bypass-first intervention, have become a more common practice at our institution throughout the study’s time period. Although this shift to minor re-interventions may seem to illuminate a short-coming of the combined MALE endpoint, this finding, especially within this subset of highly comorbid patients, may further corroborate what others have concluded, including Dick et al. following their analysis of 383 prospectively evaluated CLTI patients with diabetes undergoing either BPG-first or PTA-first revascularization: that diabetic patients with CLTI can benefit from any revascularization, regardless of the technique used; however, for some, this benefit may only occur following multiple revascularization procedures.⁴³ Although restenosis and re-intervention are valuable clinical endpoints, wound healing, MALE, amputation-free survival, and mortality are more patient-centered endpoints that were shown to not differ among treatment types. Ultimately, we feel that a bypass-first approach may provide a more durable repair in appropriately selected, anatomically suitable patients with IDDM, but the poor prognosis of patients with CLTI and diabetes warrant dedicated surveillance and individualized treatment for effective long-term success, regardless of revascularization type. Importantly, this study is not intended to fully elucidate the choice of revascularization strategy for every patient: The primary purpose of this study is to provide patients and surgeons important information regarding the outcomes of each approach, ultimately to incorporate these data and their own preferences as a means to make an effective patient-centered decision.

These data must be interpreted in the context of the study design. First, it was a retrospective, single-center review where patients were assigned to treatment based on surgeon preference, which changed over time. As our data represent the experience of one group of surgeons at a single institution, the potential for selection and information bias exists and our results are subject to the influence of specific referral patterns, surgeon experience, and patient selection preferences. Second, these data only include revascularization attempts and do not reflect outcomes for those patients treated with primary amputation or medical management. As such, we were unable to effectively collect technical success rates, as we did not capture every attempt to cross a lesion. Fortunately, several previously published studies have illustrated both the poor outcomes following medical management and the importance of revascularization in diabetic patients with CLTI.^43–47 Third, although a number of recent studies have demonstrated a benefit of drug-coated balloons and drug-eluting stents, during the time period described in this report, we had not yet widely adopted drug-eluting or –coated technology for CLTI patients.^48–50 Lastly, our analysis does not include cost-related data and therefore does not help resolve the unmet need to better understand how the high vascular readmission and revascularization rates for this prevalent and highly comorbid patient population fit into the discussion of cost-effectiveness and cost saving. Ultimately, however, our data include a large and contemporary analysis of the effect of insulin-dependence on the initial revascularization strategy for CLTI.

CONCLUSION

Our data suggest that, among patients with insulin-dependent diabetes and CLTI, a bypass first strategy is associated with similar 30-day outcomes and lower rates of both restenosis and re-intervention. These data demonstrate that a bypass-first approach may best serve appropriately selected, anatomically suitable patients with IDDM and pedal ischemia that require revascularization.