- Email: [email protected]
Infections of Prosthetic Grafts and Patches Used for Infrainguinal Arterial Reconstructions
Accepted Manuscript Infections of Prosthetic Grafts and Patches Used for Infrainguinal Arterial Reconstructions. Yana Etkin, MD, Benjamin M. Jackson, MD, Joanna S. Fishbein, MPH, Krisiva Shyta, MD, Amit Rao, MD, Hana Baig, BS, Gregg S. Landis, MD PII:
S0890-5096(18)30894-X
DOI:
https://doi.org/10.1016/j.avsg.2018.09.015
Reference:
AVSG 4132
To appear in:
Annals of Vascular Surgery
Received Date: 19 April 2018 Revised Date:
10 August 2018
Accepted Date: 9 September 2018
Please cite this article as: Etkin Y, Jackson BM, Fishbein JS, Shyta K, Rao A, Baig H, Landis GS, Infections of Prosthetic Grafts and Patches Used for Infrainguinal Arterial Reconstructions., Annals of Vascular Surgery (2018), doi: https://doi.org/10.1016/j.avsg.2018.09.015. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT 1
Title:
2
Infections of Prosthetic Grafts and Patches Used for Infrainguinal Arterial Reconstructions.
3
Authors & Affiliations:
4
Yana Etkin, MDa, Benjamin M. Jackson, MDb, Joanna S. Fishbein, MPHc, Krisiva Shyta, MDa,
5
Amit Rao, MDa, Hana Baig, BSa, Gregg S. Landis, MDa.
6
a. Division of Vascular and Endovascular Surgery, Northwell Health System, 300 Community
9
10
11
b. Division of Vascular and Endovascular Surgery, Hospital of University of Pennsylvania, 3400 Spruce Street, Philadelphia, PA, 19104 USA.
c. Biostatistics Unit, Feinstein Institute for Medical Research, Northwell Health System, 350 Community Drive, Manhasset, NY, 11030 USA.
12
Corresponding Author:
14
Yana Etkin, MD
15
1999 Marcus Ave, Suite 106B
16
Lake Success, NY 11042
17
Email: [email protected]
18
Tel: 516-233-3663
19
Fax: 516-233-3605
22 23 24 25 26 27 28
EP
AC C
21
TE D
13
20
SC
8
Drive, Manhasset, NY, 11030, USA.
M AN U
7
RI PT
1
This study was presented in the plenary session at the Eastern Vascular Society 31st Annual Meeting, Savannah, GA (October 5-8, 2017).
ACCEPTED MANUSCRIPT 2
29
Abstract:
30
Objectives: Prosthetic grafts are often used as alternative conduits in patients with peripheral
32
vascular disease who do not have an adequate autologous vein for bypass. Prosthetic grafts,
33
unfortunately, carry an increased risk of infection and are associated with increased morbidity
34
and mortality. The goal of this study was to identify potential risk factors and subsequent
35
outcomes associated with lower extremity prosthetic graft infections.
36
Methods: 272 lower extremity prosthetic bypasses and patches were performed at an academic
37
medical center between 2014 and 2016. A retrospective review of patients’ demographics, co-
38
morbidities, indication for surgery, type of procedures performed and procedural characteristics
39
was conducted. Outcomes - including limb loss and mortality - were analyzed.
40
Results: 43 patients (15.8%) with graft infections were identified during a median follow-up of
41
668 days (IQR=588). The median time to graft infection was 43 days (IQR=85) with
42
Staphylococcus being the most common bacteria cultured. Infections were associated with a
43
30.2% rate of limb loss and a 34.9% rate of mortality. The risk of infection was 2.4 times greater
44
among those with history of redo surgery [95% CI of the hazard ratio (HR): 1.3, 4.3] and 2.1
45
times greater in females (95% CI: 1.1, 3.8), by multivariable statistics. A 1 g/dL increase in
46
albumin level was associated with a 33.5% decrease in hazard of infection (HR: .67, 95% CI:
47
.46, .96) in the multivariable model. The estimated cumulative incidence of infection for female
48
patients with HTN and mean albumin of 3.36 undergoing redo surgery was 19.4% at 30 days
49
post-surgery (95% CI: 10.6, 35.6) and 39.9% at 1 year (95% CI: 26.8, 59.3).
50
Conclusions: Female gender, redo surgery and malnutrition are associated with increased risk of
51
prosthetic graft infections leading to a high rate of limb loss and mortality. Endovascular
52
interventions and bypasses with vein conduits should be considered in these patients.
AC C
EP
TE D
M AN U
SC
RI PT
31
ACCEPTED MANUSCRIPT 3
Introduction
54
Peripheral vascular disease (PVD) affects nearly 8.5 million individuals in U.S. today [1].
55
Despite advances in endovascular technologies, bypass surgery is still commonly performed for
56
limb salvage, with the autologous great saphenous vein (GSV) being the conduit of choice. Up to
57
45% [2] of patients with critical limb ischemia requiring revascularization do not have an
58
adequate continuous segment of GSV and prosthetic grafts are often used as an alternative
59
conduit in these patients.
60
Prosthetic conduits have an increased risk of infection with a reported incidence ranging from
61
.92% to 27.3% [3-7]. Multiple risk factors for graft infections have been reported including
62
surgical site infection (SSI), diabetes, female gender, presence of gangrene or active infection,
63
previous amputation, redo bypass, and early reoperation [3-6]. Graft infection is a highly morbid
64
complication with rates of amputation ranging from 8% to 52% and mortality rates of 13% to
65
58% [3].
66
Several retrospective studies analyzing lower extremity bypass infections have been previously
67
published; however, the reported data on rates of infection, outcomes and management is
68
extremely variable. This makes it challenging to draw meaningful conclusions. Several studies
69
did not describe potential risk factors for graft infections [3,7]. Others included vein bypasses in
70
the cohort, or analyzed infections only in revised bypasses [5,8]. Many studies did not
71
distinguish between SSI and graft infection [8,9]. In other reports, potential risk factors for graft
72
infection were identified by analyzing only patients who developed infections rather than
73
comparing them to cohort without infection [3,6].
AC C
EP
TE D
M AN U
SC
RI PT
53
ACCEPTED MANUSCRIPT 4
The purpose of our study was to determine the risk factors associated with infection of prosthetic
75
conduits used for lower extremity arterial reconstruction and to analyze the effect these
76
infections have on patient outcomes.
77
Methods
78
We performed a retrospective review of all patients undergoing lower extremity arterial
79
reconstruction with prosthetic grafts or patches at a tertiary academic medical center between
80
January 2014 and December 2016. Patient characteristics, indication for intervention, type of
81
reconstruction performed, and procedural characteristics were recorded. Incidence of graft
82
infection was the primary end point of the study. The rates of limb loss and mortality during the
83
follow up period were analyzed as the secondary end points. Data collection and analysis were
84
conducted in accordance with the Northwell Health Institutional Review Board (IRB). The
85
waiver of informed consent and Health Insurance Portability and Accountability Act waiver of
86
authorization were approved by the IRB committee.
87
Polytetrafluoroethylene (PTFE) grafts or Dacron patches were used in all patients included in the
88
study. Those undergoing creation of upper extremity bypasses, dialysis grafts, or implantation of
89
bovine pericardium patches were excluded. All the patches and at least one of the anastomotic
90
sites for the bypass grafts were located in the infrainguinal region. Index procedure was
91
considered to be a redo surgery if patients had previously undergone a different open
92
revascularization in the ipsilateral limb. Any open surgical interventions after index
93
revascularizations were recorded as bypass revisions.
AC C
EP
TE D
M AN U
SC
RI PT
74
94
A graft infection was defined based on the presence of infected fluid or inflammation directly
95
communicating with the prosthetic material, seen on imaging or intraoperatively. Infection was
ACCEPTED MANUSCRIPT 5
confirmed by positive culture of surrounding fluid or explanted graft. All exposed grafts were
97
considered to be infected as well. Patients with isolated SSIs were excluded from the final
98
cohort.
99
Our protocol is to administer antibiotics 60 minutes before skin incision. The choice of
100
antibiotics and type of skin preparation solution used was surgeon dependent. The draping of the
101
surgical field, including preparation and isolation of infected/gangrenous tissues, were not
102
standardized. Ioban use was surgeon specific as well. Electrical clippers were used for hair
103
removal in all cases. Preoperative hibiclens was not used routinely. Incisional closure was not
104
standardized and negative pressure incision management devices, such as PREVENA™, were
105
not utilized. The management of foot infection or gangrene present at the time of
106
revascularization was patient and surgeon specific. Most surgeons in our group prefer to perform
107
wound debridement and/or toe amputations as a separate procedure several days after the
108
revascularization. Standard follow up after lower extremity revascularization at our institution is
109
one and three months postoperatively, then six months thereafter. The social security death index
110
and hospital electronic medical records were used to record survival during the follow-up period.
111
Limb loss was defined as either below knee (BKA) or above knee amputation (AKA).
112
Summary statistics (e.g., mean and standard deviation, or median and interquartile range) were
113
calculated for continuous variables as well as frequencies and proportions for categorical
114
variables. The data for patient with grafts and patches was combined for this analysis. Subgroup
115
analysis was not performed due to the relatively small number of patients in our cohort.
116
Univariate and multivariable survival analysis methods were used to assess the association
117
between potential factors and risk of graft infection. Specifically, the Fine and Gray method for
118
competing-risks analysis was performed, whereby death was considered the competing event,
AC C
EP
TE D
M AN U
SC
RI PT
96
ACCEPTED MANUSCRIPT 6
and the event of interest is first instance of graft infection. Time-to-event was measured as the
120
number of days from surgery to graft infection (or mortality). Univariate subdistribution hazards
121
regression for competing risks was used for continuous variables. Factors that appeared to be
122
significantly associated with survival in the univariate analysis at alpha level 0.1 (including the
123
multiple comparisons tests), were considered for inclusion in the subdistribution hazards
124
regression multivariable model. Less strict p-value criteria was used during univariate analyses
125
to allow for identifying variables that may be confounded by another variable. Final model was
126
selected using the likelihood ratio test.
127
Cox regression extended for time-dependent covariates was performed to assess if presence of
128
graft infection influenced patient overall survival or limb loss where mortality was treated as
129
competing risk and graft infection status as time-varying covariate. A result was considered
130
significant at the P<.05 level of significance, unless otherwise stated. All analyses were
131
conducted using SAS version 9.4 (SAS Institute Inc., Cary, NC).
132
Results
133
A total of 272 patients underwent lower extremity reconstructions with 228 PTFE bypass grafts
134
and 44 Dacron patches. Of these, 58.8% were male with an average age of 72.6 ± 10.7 years.
135
64.3% of subjects were Caucasian. Median follow-up time was 668 days (IQR=588), ranging
136
from 0 to 1,363 days. Thirty-seven bypass grafts (16.2%) and six patches (13.6%) were found to
137
be infected. The overall rate of graft infection was 15.8% (43/272) with median time to
138
presentation of 43 days (IQE=85) after initial revascularization. The incidence of superficial
139
surgical site infection was 20.2% (55/272). Thirty five patients with graft infections had
140
concomitant superficial SSIs. The other 20 patients did not develop graft infections despite
141
documented signs of wound infection.
AC C
EP
TE D
M AN U
SC
RI PT
119
ACCEPTED MANUSCRIPT 7
Univariate competing risks analysis comparing patients with and without infection identified
143
female gender (P=.01), hypertension (P=.08), serum albumin level (P=.02), and redo surgery
144
(P=.01) as potential predictors of infection. No other comorbidities or patient characteristics were
145
associated with risk of infection, including race, age, insurance status, body mass index (BMI),
146
diabetes, heart disease, lung disease, renal insufficiency or active smoking (Table I). Presence of
147
acute limb ischemia and need for emergency surgery were not predictive of infection. The
148
presence of open wounds or gangrene on the operative limb was also not significantly associated
149
with the outcome (Table II). The type of revascularization performed or procedural
150
characteristics were not associated with infection either. The length of hospital stay prior to
151
procedure was similar in patients with and without infection. (Tables III and IV).
152
Average length of surgery was similar in both groups (238 min vs. 256 min, P=.26).
153
Perioperative antibiotics were administered appropriately in all the cases and Ancef was used
154
62.7% of the time. Chlorhexidine gluconate with isopropyl alcohol was used for skin
155
preparation in 74.0% of the cases. The rate and volume of intraoperative blood transfusions did
156
not correlate with risk of infection. In the group of patients without infections, 35.4% of them
157
received transfusions at an average of 2.2 units of blood per person. Similarly, in the graft
158
infection group, 37.2% of patients had transfusions at an average of 1.7 units per person.
159
Eighteen bypasses underwent revisions after the index procedure with similar rates in both
160
groups, 5.7% vs. 11.6% (P=.18).
161
The final multivariable model included redo surgery status, hypertension, gender, and albumin
162
level. The analysis suggests that the hazard for graft infection is 2.4 times greater among those
163
with a history of redo of surgery (95% CI: 1.3, 4.3, P=.01) and 2.1 times greater in females
164
compared to males (95% CI: 1.1, 3.8, P=.02) after adjusting for other factors in the model. A 1-
AC C
EP
TE D
M AN U
SC
RI PT
142
ACCEPTED MANUSCRIPT 8
unit increase in albumin level is associated with a 33.5% decrease in risk of infection (HR: .7,
166
95% CI: .5, .9, P=.03). Cumulative incidence was computed, holding albumin at the mean of
167
3.36 g/dl and comparing gender, hypertension status and history of redo surgery. The highest
168
incidence of infection was seen among female patients with hypertension undergoing redo
169
operation (Figure 1). For instance, estimated cumulative incidence of infection in this group was
170
19.4% at 30 days post-surgery (95% CI: 10.6, 35.6) and 39.9% at 1 year (95% CI: 26.8, 59.3).
171
Thirteen out of forty-three subjects (30.2%) suffered limb loss due to infection compared to the
172
8.7% (20/229) rate of amputation in patients without infections. The risk of limb loss was 8.1
173
times greater for patients with infections (95% CI for HR: 3.5, 18.5, P<.0001). Graft infection,
174
modeled as a time-dependent covariate, was also found to be a significant predictor of overall
175
survival. The rate of mortality during the follow period was 34.9% (15/43) in the infection group
176
vs. 13.1% (30/229) without infection. The risk of death was 6.4 times greater for a subject with
177
graft infection compared to one without infection (95% CI: 3.3, 12.3, P<.0001).
178
We employed several treatment options to manage bypass/patch infections, which was physician
179
and patient dependent (Table V). Thirty-two (76.2%) of the infected bypasses were patent on
180
presentation. Graft excision with or without reconstruction was performed in 53.5% of the cases,
181
while graft preservation was attempted in 39.5% of the cases. Graft preservation involved
182
aggressive wound debridement, followed by muscle flap coverage in 70.6% of the cases.
183
Vacuum assisted closure (VAC) devices were used in all the patients. Three patients were
184
severely sick due to infection and elected palliative care measures. The rate of limb loss was
185
significantly higher in the graft excision group as compared to the graft preservation group,
186
52.2% vs. 5.9% (P=.002). Mortality rates, however, were not affected by a chosen treatment
187
(30.4% vs 29.4%, P=.62). Staphylococcus was the most common bacteria, isolated in 51.2% of
AC C
EP
TE D
M AN U
SC
RI PT
165
ACCEPTED MANUSCRIPT 9
the cultures. Of those, 36.4% were methicillin-resistant Staphylococcus aureus (MRSA) (Table
189
VI). There was no correlation between culture results and patient outcomes.
190
Discussion
191
Surgical site infections (SSI) after lower extremity revascularization are relatively common and
192
well described in the literature, with reported incidence ranging from 4.8% to 15.6% [10-12].
193
Common factors associated with SSI have been identified in several recent studies, including the
194
data from Vascular Quality Initiative registry and American College of Surgeons National
195
Surgical Quality Improvement Program [10,11]. Transfusion rate, procedure time, skin
196
preparation, female gender, obesity, dialysis, and postoperative seroma are some of the common
197
factors associated with increased risk of SSI [10,11,13,14]. These superficial infections have
198
been shown to lead to deep infections involving prosthetic graft material used for
199
revascularization. As compared to SSI data, the literature on incidence, risk factors, and
200
outcomes of graft infections is less robust. No large prospective studies referring to these
201
variables are available at this time. Multiple previous studies did not distinguish wound
202
infections from graft infections, making it hard to interpret the results [8,9,15]. In our study,
203
twenty patients with wound infection did not develop graft infection, underscoring the fact that
204
these complications should be analyzed separately. Determining risks factors specific to graft
205
infection may allow us to reduce the rate of this troubling complication associated with high
206
morbidity and mortality rates.
207
During the three-year study period, we observed a 15.8% rate of prosthetic graft infections,
208
which is similar to previously published data [3-7]. We analyzed 44 patients with infections,
209
which is one of the largest cohort of patients reported in the literature to date. Female gender,
AC C
EP
TE D
M AN U
SC
RI PT
188
ACCEPTED MANUSCRIPT 10
210
redo surgery and low albumin were significant predictors of graft/patch infection in our
211
population. Our study suggests that females have more than a two-fold increase risk of infection as
213
compared to males. Similarly, Siracuse et al. reported 4.5 times higher rates of graft infection in
214
females [6]. This gender-related difference cannot be attributed to body weight alone, as BMI did
215
not correlate with infection in our study. Women have been shown to have greater lower
216
extremity fat distribution as compared to men [6,11,16], and may explain higher rates of
217
infection after lower extremity operations.
218
The association of redo surgery with prosthetic graft infection was also previously described in
219
the literature [3,6]. Often, these are technically challenging operations with prolonged dissection
220
time. Increased risk of infection may be due to poor vascularization and scarring encountered in
221
the re-operative field. As demonstrated by several studies, prolonged operative time correlates
222
with inferior healing rates and an increase in SSI [3,8,11,14].
223
It is not surprising that low albumin level in our study was associated with an increased risk of
224
infection. It is well established that hypoalbuminemia, as a marker of malnutrition, is associated
225
with a greater risk of adverse surgical outcomes. Results of the National VA Surgical Risk study
226
demonstrated that 1 g/dl decrease in albumin level was associated with more than a 2-fold
227
increase in mortality and morbidity. In the multivariate analyses, albumin level was a strong
228
predictor of major infection complications including systemic sepsis, pneumonia, and deep
229
wound infection [17]. An albumin level of less than 3.5 g/dl was found to be one of the fourteen
230
independent predictors of postoperative surgical site infection after general and vascular surgery,
231
based on results of the NSQIP patient safety in surgery study [18]. Several other factors that
AC C
EP
TE D
M AN U
SC
RI PT
212
ACCEPTED MANUSCRIPT 11
were not analyzed in our study have been shown to be more accurate predictors of nutritional
233
status, such as prealbumin, transferrin and retinol-binding protein levels, as well nutritional
234
assessment based on history and physical examination [19].
235
Prior studies demonstrated other potential risk factors for graft infection not observed in our
236
population. Early graft revision was reported as a predictor of prosthetic graft infections in
237
several studies. Brothers et al. described an eleven-fold increase in the risk of graft infection after
238
a bypass revision [4]. Kolakowski et al. reported an 11% graft infection rate in patients who had
239
a revision <30 days after the initial bypass operation versus 6.1% in those whose revision took
240
place >30 days later [5]. Based on our analysis, rates of bypass revision were not significantly
241
different in patient with graft infection as compared to those without infection. Diabetes, active
242
infection, limb amputation, and prolonged operative time, are other reported predictors for the
243
development of graft infection that were not observed in our study [3,4,6,8].
244
Management of infected prosthetic material continues to evolve. Complete excision of an
245
infected graft is still the preferred method of treatment; however, it places patients at a greater
246
risk for limb loss. Arterial reconstruction with extra-anatomic or in-situ graft replacement have
247
been described utilizing various conduits including biosynthetic prosthesis, cryopreserved human
248
allograft, prosthetic or a vein grafts [7,20-23]. In our study, more than two thirds of infected
249
bypasses were patent on presentation and graft preservation was attempted in 39.5% of these
250
cases. Mortality rates did not differ; however, rates of amputation were significantly lower in the
251
graft preservation group. The rates of repeated intervention and continued sepsis in the graft
252
preservation group was not recorded in our study, but likely to be higher, as reported by other
253
groups. Martens and colleagues described an increased incidence of subsequent operations for
254
continued sepsis in patient treated with incomplete removal of infected grafts; 82% vs. 13% [7].
AC C
EP
TE D
M AN U
SC
RI PT
232
ACCEPTED MANUSCRIPT 12
Several successful techniques of graft preservation have been described in the literature
256
including muscle flap coverage of the infected graft, and vacuum assisted closure. Morasch et al.
257
demonstrated an 85% rate of prosthetic graft salvage with the use of a gracilis muscle flap [24].
258
Similarly, coverage with a sartorius muscle flap was successful in 86% of cases reported by
259
Landry and colleagues [25]. Several other series described debridement and vacuum assisted
260
closure with rates of graft preservation ranging from 76% to 83% of patients [9,26]. Antibiotic-
261
loaded polymethyl methacrylate (PMMA) beads is another adjunct that can be used to improve
262
graft salvage rates [27]. Multiple series published in literature demonstrate high rates of
263
successful limb salvage utilizing these alternative treatment options. However, some
264
contradictory reports have been published. Taylor et al. described 78% rate of recurrent infection
265
after local debridement and muscle coverage was attempted [28]. At this time, it’s difficult to
266
draw definitive conclusions on outcomes of graft preservation techniques based on small,
267
retrospective series without long term follow up results.
268
Graft infections continue to be a major complication of revascularization procedures, particularly
269
those involving prosthetic material. In our population, these infections were associated with 30%
270
rate of limb loss and 35% mortality. These rates have not improved as compared to studies
271
published several decades ago [3,7,20]. Given the increased mortality and morbidity these
272
infections place on already vulnerable population, it is important to consider other
273
revascularization options and alternative conduits, when appropriate. Endovascular options, such
274
as lower extremity stents and angioplasty are worth considering. In the presence of diffuse
275
vascular disease, a bypass may be the only viable choice. The use of GSV as a bypass conduit
276
remains the gold standard, however, in the absence of an adequate GSV it may be worthwhile to
277
consider the use of other native veins, as autologous conduits demonstrate lower infection rates.
AC C
EP
TE D
M AN U
SC
RI PT
255
ACCEPTED MANUSCRIPT 13
278
Lloyd and colleagues demonstrated similar patency rates between ipsilateral GSV to opposite leg
279
veins, arm veins, lesser saphenous veins, superficial femoral veins, and spliced veins when used
280
as
281
This study is limited by its single center retrospective review design. The database is not entirely
282
comprehensive and certain parameters that may have potential impact on infection rates were not
283
evaluated, such as HbA1c, preoperative medication, WiFi classification of tissue loss, nutritional
284
parameters, etc. Similarly, we cannot account for some important procedural details such as
285
preparation and draping of the surgical field, management of infected or gangrenous tissues, and
286
skin closure techniques. There are several potential confounders outside our study design that we
287
cannot account for, such as technical skill of the surgeons, intraoperative management and
288
anesthesia, as well as postoperative incisional care.
289
Conclusion
290
Female gender, redo surgery and malnutrition are associated with increased risk of prosthetic
291
graft infections leading to high rates of limb loss and mortality. Endovascular interventions and
292
bypasses with vein conduits should be considered in these patients.
conduits
[2].
AC C
EP
TE D
M AN U
SC
RI PT
alternative
ACCEPTED MANUSCRIPT 1
References 1. Reed JF, Eid S, Edris B, et al. Prevalence of peripheral artery disease varies significantly
Rehabil 2009;16:377-81. doi:10.1097/hjr.0b013e32832955e2.
RI PT
depending upon the method of calculating ankle brachial index. Eur J Cardiovasc Prev
2. Taylor LM, Edwards JM, Brant E, et al. Autogenous reversed vein bypass for lower
1987;153:505–10. doi:10.1016/0002-9610(87)90803-8.
SC
extremity ischemia in patients with absent or inadequate greater saphenous vein. Am J Surg
M AN U
3. Edwards WH Jr, Martin RS, Jenkins JM, et al. Primary graft infections. J Vasc Surg 1987;6:235-9. doi:10.1016/0741-5214(87)90034-6.
4. Brothers T, Robison J, Elliott B. Predictors of Prosthetic Graft Infection after Infrainguinal Bypass. J Am Coll Surg 2009;208:557-61. doi:10.1016/j.jamcollsurg.2009.01.001. 5. Kolakowski S Jr, Dougherty MJ, Calligro KD. Does the timing of reoperation influence the
TE D
risk of graft infections? J Vac Surg 2007;45:60-4. doi:10.1016/j.jvs.2006.09.007. 6. Siracuse JJ, Nandiva P, Giles KA, et al. Ten-year experience with prosthetic graft infections involving the femoral artery. J Vasc Surg 2013;57:700-5. doi:10.1016/j.jvs.2012.09.049.
EP
7. Mertens RA, O’Hara PJ, Hertzer NR, et al. Surgical management of infrainguinal arterial
AC C
prosthetic graft infections: review of a thirty-five-year experience. J Vasc Surg 1995;21:78290.
8. Chang JK, Calligaro KD, Ryan S, et al. Risk factors associated with infection of lower extremity revascularization: analysis of 365 procedures performed at a teaching hospital. Ann Vasc Surg 2003;17:91-6. doi:10.1007/s10016-001-0337-8.
ACCEPTED MANUSCRIPT 2
9. Dosluoglu HH, Loghmanee C, Lall P, et al. Management of early (<30 day) vascular groin infection using vacuum-assisted closure alone without muscle flap coverage in consecutive patient series. J Vasc Surg 2010;51:1160-6. doi:10.1016/j.jvs.2009.11.053.
RI PT
10. Kalish JA, Farber A, Homa K, et al. Factors associated with surgical site infection after lower extremity bypass in the Society for Vascular Surgery (SVS) Vascular Quality Initiative (VQI). J Vasc Surg 2014;60:1238-46. doi:10.1016/j.jvs.2014.05.012.
lower
extremity
revascularization.
J
Vasc
Surg
2011;54:433–9.
M AN U
doi:10.1016/j.jvs.2011.01.034.
SC
11. Greenblatt DY, Rajamanickam V, Mell MW. Predictors of surgical site infection after open
12. Adam DJ, Beard JD, Cleveland T, et al. Bypass versus angioplasty in severe ischemia of the leg (BASIL): multicenter, randomized controlled trial. Lancet 2005;366:1925–34. doi:10.1016/j.jvs.2010.01.077.
procedures
TE D
13. Kuy S, Dua A, Desai S, et al. Surgical site infections after lower extremity revascularization involving
groin
incisions.
Ann
Vasc
Surg
2014;28:53-8.
doi:10.1016/j.avsg.2013.08.002.
EP
14. Davis FM, Sutzko DC, Grey SF, et al. Predictors of surgical site infection after open lower extremity revascularization. J Vasc Surg 2017;65:1769-78. doi:10.1016/j.jvs.2016.11.053.
AC C
15. Jensen LJ, Kimose HH. Prosthetic graft infections: a review of 720 arterial prosthetic reconstructions. Thorac Cardiovasc Surg 1985;33:391-8. doi:10.1055/s-2007-1014176. 16. Nguyen LL, Brahmanandam S, Bandyk DF, et al. Female gender and oral anticoagulants are associated with wound complications in lower extremity vein bypass: an analysis of 1404 operations
for
critical
doi:10.1016/j.jvs.2007.07.053.
limb
ischemia.
J
Vasc
Surg
2007;46:1191–7.
ACCEPTED MANUSCRIPT 3
17. Gibbs J, Cul W, Henderson W, et al. Preoperative Serum Albumin Level as a Predictor of Operative Mortality and Morbidity. Arch Surg 1999;134:36-42. 18. Neumayer L, Hosokawa P, Itani K, et al. Multivariable Predictors of Postoperative Surgical
RI PT
Site Infection after General and Vascular Surgery: Results from the Patient Safety in Surgery Study. J Am Coll Surg 2007;204:1178-87. doi:10.1016/j.jamcollsurg.2007.03.022.
19. Bharadwaj S, Ginoya S, Tandon P, et al. Malnutrition: laboratory markers vs. nutritional
SC
assessment. Gastroenterology Report 2016;4:272-80. doi:10.1093/gastro/gow013.
infection. AM Surg 1998;645:39-45.
M AN U
20. Henke PK, Bergamini TM, Rose SM, et al. Current options in prosthetic vascular graft
21. Topel I, Betz T, Uhl C, et al. Use of biosynthetic prosthesis to replace infected infrainguinal prosthetic grafts-- first results. VASA 2012;41:212-20. doi:10.1024/0301-1526/a000188. 22. Wiltberg G, Matia L, Schmelzle M, et al. Mid- and long-term results after replacement of
TE D
infected peripheral vascular prosthetic grafts with biosynthetic collagen prosthesis. J Cardiovasc Surg 2014;55:693-8.
23. Brown KE, Heyer K, Rodriguez H, et al. Arterial reconstruction with cryopreserved human
EP
allograft in the setting of infection: a single-center experience with midterm follow-up. J Vasc Surg 2009;49:660-6. doi:10.1016/j.jvs.2008.10.026.
AC C
24. Morasch MD, Sam AD, Kibbe MR, et al. Early results with use of gracilis muscle flap coverage of infected groin wound after vascular surgery. J Vasc Surg 2004;39:1277-83. doi:10.1016/j.jvs.2004.02.011. 25. Landy GJ, Carlson JR, Liem TK, et al. The sartorius muscle flap: an important adjunct for complicated femoral wounds involving vascular grafts. Am J Surg 2009;197:655-9. doi:10.1016/j.amjsurg.2008.12.020.
ACCEPTED MANUSCRIPT 4
26. Acosta S, Monsen C. Outcome after VAC therapy for infected bypass grafts in the lower limb. Eur J Vasc Endovasc Surg 2012;44:294-9. doi:10.1016/j.ejvs.2012.06.005.
RI PT
27. Stone PA, Mousa AY, Hass SM, et al. Antibiotic-loaded polymethylmethacrylate beads for treatment of extracavitary vascular surgical site infections. J Vasc Surg 2012;55:1706-11. doi:10.1016/j.jvs.2011.12.037.
SC
28. Taylor SM, Weatherford DA., Langan EM, et al. Outcomes in the management of vascular prosthetic graft infection confined to the groin: a reappraisal. Ann Vasc Surg 1996; 10:117-
AC C
EP
TE D
M AN U
22. doi:10.1007/BF02000754.
ACCEPTED MANUSCRIPT Characteristics
Patients without infection N= 229 n (%)
Patients with infection N = 43 n (%)
SHR with 95% CI, P-value*
72.2 ±10.9
74.6 ± 9.07
1.0 (.09-1.0), .12
87 (38.0)
25 (58.1)
2.1 (1.2-3.9), .01
Caucasian
148 (64.6)
29 (67.4)
n/a
African American
46 (20.1)
10 (23.3)
1.1 (.5-2.3), .75
Asian
17 (7.4)
1 (2.3)
.3 (.05-2.2), .25
Hispanic
5 (2.2)
Age (years), mean ±SD Sex (% female)
SC
RI PT
Race
3 (7)
Medicare
170 (74.2)
Medicaid
30 (13.1)
Private
29 (12.7)
M AN U
Insurance
.8 (.2-3.3), .73
36 (83.7)
1.5 (.5-4.2), .46
3 (7)
.7 (.2-3.1), .65
4 (9.3)
n/a
26 (60.5)
1.6 (.9-2.9), .14
194 (84.7)
41 (95.4)
3.5 (.8-14.2), .08
96 (41.9)
17 (39.5)
.9 (.5-1.7), .8
28 (12.2)
9 (20.9)
1.8 (.8-3.7), .13
45 (19.7)
8 (18.6)
.4 (.1-1.6), .2
42 (18.3)
9 (20.9)
1.1 (.5-2.2), .8
BMI (kg/m2), mean ±SD
26.9 ±5.6
28.2 ± 5.2
1.0 (1.0-1.1), .13
Albumin (g/dl), mean ±SD
3.4 ±0.7
3.1 ± 0.7
.64 (.4-.9), .02
111 (48.5)
Hypertension Coronary artery disease
EP
COPD
TE D
Diabetes Mellitus
Renal Disease
AC C
Current Smoker
*The Fine and Gray proportional subdistribution hazard model was used with all-cause death as a competing risk.
ACCEPTED MANUSCRIPT
Patients with infection N = 43 n (%)
SHR with 95% CI, P-value*
Claudication
31 (13.5)
4 (9.3)
.9 (.2-3.4), .84
Rest pain
39 (17.0)
8 (18.6)
1.3 (.4-3.9), .68
Open Wound
55 (24.0)
13 (30.2)
1.4 (1.4-.5), .5
Gangrene
47 (20.5)
10 (23.3)
1.3 (1.3-.4), .61
23 (10)
3 (7)
.7 (.2-2.1), .5
Aneurysms Acute limb ischemia
RI PT
Patients without infection N = 229 n (%)
SC
Indication for Surgery
34 (14.9)
5 (11.6)
.8 (.3-2.1), .62
M AN U
*The Fine and Gray proportional subdistribution hazard model was used with all-cause death as
AC C
EP
TE D
a competing risk.
ACCEPTED MANUSCRIPT
Type of Surgery
Patients without infection N = 229 n (%)
Patients with infection N = 43 n (%)
Aorta-Femoral
7 (3.1)
0
Ilio-Femoral
19 (8.3)
2 (4.7)
Axillary-Femoral
11 (4.8)
5 (11.6)
SHR with 95% CI, P-value*
Femoral-Femoral
37 (16.2)
4 (9.3)
Femoral-Popliteal
79 (34.5)
14 (32.6)
RI PT
.3 (.1-1.2), .09
Femoral-Tibial
38 (16.6)
12 (27.9)
n/a
SC
.6 (.3-1.5), .32
.6 (.3-1.3), .19
M AN U
Femoral 38 (16.6) 6 (14) .6 (.2-1.5), .26 Endarterectomy *The Fine and Gray proportional subdistribution hazard model was used with all-cause death as
AC C
EP
TE D
a competing risk.
Patients without infection N = 229 n (%)
Patients with infection N = 43 n (%)
SHR with 95% CI, P-value*
142 (62.0)
26 (60.5)
1.0 (.5-1.9), .95
Vancomycin
34(14.8)
6 (14.0)
n/a
Other
53 (23.2)
11 (25.5)
n/a
Chlorhexidine gluconate
170 (74.2)
29 (67.4)
n/a
Betadine
59 (25.8)
SC
ACCEPTED MANUSCRIPT Characteristics
14 (32.6)
1.2 (.6-2.3), .6
Procedure Time (min), mean ±SD
238 ± 113
256 ± 130
1.1 (.9-1.2), .26
Rate of blood transfusions
81 (35.4)
16 (37.2)
.9 (.8-1.1), .5
Time from admission to surgery(d), median(range)
3.0 (0-8)
1.0 (0-7)
1.0 (1.0-1.1), .19
Redo surgery
49 (21.4)
17 (39.5)
2.2 (1.3-4.3), .01
13 (5.7)
5 (11.6)
1.8 (.8-4.1), .18
Ancef
TE D
Bypass Revision
M AN U
Skin preparation
RI PT
Perioperative Antibiotics
AC C
EP
*The Fine and Gray proportional subdistribution hazard model was used with all-cause death as a competing risk.
ACCEPTED MANUSCRIPT
Limb Loss n(%)
Mortality n(%)
Graft removal ± revascularization
23 (53.5)
12 (52.2)
7 (30.4)
No revascularization
15 (65.2)
12 (80)
5 (33.3)
In-situ bypass with vein
4 (17.4)
0
In-situ bypass with homograft
2 (8.7)
0
Extra-anatomic bypass
2 (8.7)
0
Graft preservation
17 (39.5)
RI PT
n(%)
SC
Treatment (N=43)
2 (50.0) 0 0
1 (5.9)
5 (29.4)
1 (5.9)
3 (25.0)
12 (70.6)
Debridement + VAC
5 (29.4)
0
2 (40.0)
3 (7)
0
3 (100)
M AN U
Debridement + muscle flap+ VAC
AC C
EP
TE D
Palliative care
ACCEPTED MANUSCRIPT
All Staphylococcus species
N=43 (%) 51.2
Methicillin- Resistant Staphylococcus aureus
18.6
Enterococcus
20.9
RI PT
Bacteria Isolated
Pseudomonas aeruginosa
18.6
Proteus
16.3 11.6
SC
E.coli Enterobacter
4.7
M AN U
Yeast Streptococcus Klebsiella pneumoniae Morganella morganii
Corynebacterium Polymicrobial
AC C
EP
No growth
TE D
Acenetobacter
4.7 2.3 2.3 2.3 2.3 2.3
34.9 4.7
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
S0890-5096(18)30894-X
DOI:
https://doi.org/10.1016/j.avsg.2018.09.015
Reference:
AVSG 4132
To appear in:
Annals of Vascular Surgery
Received Date: 19 April 2018 Revised Date:
10 August 2018
Accepted Date: 9 September 2018
Please cite this article as: Etkin Y, Jackson BM, Fishbein JS, Shyta K, Rao A, Baig H, Landis GS, Infections of Prosthetic Grafts and Patches Used for Infrainguinal Arterial Reconstructions., Annals of Vascular Surgery (2018), doi: https://doi.org/10.1016/j.avsg.2018.09.015. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT 1
Title:
2
Infections of Prosthetic Grafts and Patches Used for Infrainguinal Arterial Reconstructions.
3
Authors & Affiliations:
4
Yana Etkin, MDa, Benjamin M. Jackson, MDb, Joanna S. Fishbein, MPHc, Krisiva Shyta, MDa,
5
Amit Rao, MDa, Hana Baig, BSa, Gregg S. Landis, MDa.
6
a. Division of Vascular and Endovascular Surgery, Northwell Health System, 300 Community
9
10
11
b. Division of Vascular and Endovascular Surgery, Hospital of University of Pennsylvania, 3400 Spruce Street, Philadelphia, PA, 19104 USA.
c. Biostatistics Unit, Feinstein Institute for Medical Research, Northwell Health System, 350 Community Drive, Manhasset, NY, 11030 USA.
12
Corresponding Author:
14
Yana Etkin, MD
15
1999 Marcus Ave, Suite 106B
16
Lake Success, NY 11042
17
Email: [email protected]
18
Tel: 516-233-3663
19
Fax: 516-233-3605
22 23 24 25 26 27 28
EP
AC C
21
TE D
13
20
SC
8
Drive, Manhasset, NY, 11030, USA.
M AN U
7
RI PT
1
This study was presented in the plenary session at the Eastern Vascular Society 31st Annual Meeting, Savannah, GA (October 5-8, 2017).
ACCEPTED MANUSCRIPT 2
29
Abstract:
30
Objectives: Prosthetic grafts are often used as alternative conduits in patients with peripheral
32
vascular disease who do not have an adequate autologous vein for bypass. Prosthetic grafts,
33
unfortunately, carry an increased risk of infection and are associated with increased morbidity
34
and mortality. The goal of this study was to identify potential risk factors and subsequent
35
outcomes associated with lower extremity prosthetic graft infections.
36
Methods: 272 lower extremity prosthetic bypasses and patches were performed at an academic
37
medical center between 2014 and 2016. A retrospective review of patients’ demographics, co-
38
morbidities, indication for surgery, type of procedures performed and procedural characteristics
39
was conducted. Outcomes - including limb loss and mortality - were analyzed.
40
Results: 43 patients (15.8%) with graft infections were identified during a median follow-up of
41
668 days (IQR=588). The median time to graft infection was 43 days (IQR=85) with
42
Staphylococcus being the most common bacteria cultured. Infections were associated with a
43
30.2% rate of limb loss and a 34.9% rate of mortality. The risk of infection was 2.4 times greater
44
among those with history of redo surgery [95% CI of the hazard ratio (HR): 1.3, 4.3] and 2.1
45
times greater in females (95% CI: 1.1, 3.8), by multivariable statistics. A 1 g/dL increase in
46
albumin level was associated with a 33.5% decrease in hazard of infection (HR: .67, 95% CI:
47
.46, .96) in the multivariable model. The estimated cumulative incidence of infection for female
48
patients with HTN and mean albumin of 3.36 undergoing redo surgery was 19.4% at 30 days
49
post-surgery (95% CI: 10.6, 35.6) and 39.9% at 1 year (95% CI: 26.8, 59.3).
50
Conclusions: Female gender, redo surgery and malnutrition are associated with increased risk of
51
prosthetic graft infections leading to a high rate of limb loss and mortality. Endovascular
52
interventions and bypasses with vein conduits should be considered in these patients.
AC C
EP
TE D
M AN U
SC
RI PT
31
ACCEPTED MANUSCRIPT 3
Introduction
54
Peripheral vascular disease (PVD) affects nearly 8.5 million individuals in U.S. today [1].
55
Despite advances in endovascular technologies, bypass surgery is still commonly performed for
56
limb salvage, with the autologous great saphenous vein (GSV) being the conduit of choice. Up to
57
45% [2] of patients with critical limb ischemia requiring revascularization do not have an
58
adequate continuous segment of GSV and prosthetic grafts are often used as an alternative
59
conduit in these patients.
60
Prosthetic conduits have an increased risk of infection with a reported incidence ranging from
61
.92% to 27.3% [3-7]. Multiple risk factors for graft infections have been reported including
62
surgical site infection (SSI), diabetes, female gender, presence of gangrene or active infection,
63
previous amputation, redo bypass, and early reoperation [3-6]. Graft infection is a highly morbid
64
complication with rates of amputation ranging from 8% to 52% and mortality rates of 13% to
65
58% [3].
66
Several retrospective studies analyzing lower extremity bypass infections have been previously
67
published; however, the reported data on rates of infection, outcomes and management is
68
extremely variable. This makes it challenging to draw meaningful conclusions. Several studies
69
did not describe potential risk factors for graft infections [3,7]. Others included vein bypasses in
70
the cohort, or analyzed infections only in revised bypasses [5,8]. Many studies did not
71
distinguish between SSI and graft infection [8,9]. In other reports, potential risk factors for graft
72
infection were identified by analyzing only patients who developed infections rather than
73
comparing them to cohort without infection [3,6].
AC C
EP
TE D
M AN U
SC
RI PT
53
ACCEPTED MANUSCRIPT 4
The purpose of our study was to determine the risk factors associated with infection of prosthetic
75
conduits used for lower extremity arterial reconstruction and to analyze the effect these
76
infections have on patient outcomes.
77
Methods
78
We performed a retrospective review of all patients undergoing lower extremity arterial
79
reconstruction with prosthetic grafts or patches at a tertiary academic medical center between
80
January 2014 and December 2016. Patient characteristics, indication for intervention, type of
81
reconstruction performed, and procedural characteristics were recorded. Incidence of graft
82
infection was the primary end point of the study. The rates of limb loss and mortality during the
83
follow up period were analyzed as the secondary end points. Data collection and analysis were
84
conducted in accordance with the Northwell Health Institutional Review Board (IRB). The
85
waiver of informed consent and Health Insurance Portability and Accountability Act waiver of
86
authorization were approved by the IRB committee.
87
Polytetrafluoroethylene (PTFE) grafts or Dacron patches were used in all patients included in the
88
study. Those undergoing creation of upper extremity bypasses, dialysis grafts, or implantation of
89
bovine pericardium patches were excluded. All the patches and at least one of the anastomotic
90
sites for the bypass grafts were located in the infrainguinal region. Index procedure was
91
considered to be a redo surgery if patients had previously undergone a different open
92
revascularization in the ipsilateral limb. Any open surgical interventions after index
93
revascularizations were recorded as bypass revisions.
AC C
EP
TE D
M AN U
SC
RI PT
74
94
A graft infection was defined based on the presence of infected fluid or inflammation directly
95
communicating with the prosthetic material, seen on imaging or intraoperatively. Infection was
ACCEPTED MANUSCRIPT 5
confirmed by positive culture of surrounding fluid or explanted graft. All exposed grafts were
97
considered to be infected as well. Patients with isolated SSIs were excluded from the final
98
cohort.
99
Our protocol is to administer antibiotics 60 minutes before skin incision. The choice of
100
antibiotics and type of skin preparation solution used was surgeon dependent. The draping of the
101
surgical field, including preparation and isolation of infected/gangrenous tissues, were not
102
standardized. Ioban use was surgeon specific as well. Electrical clippers were used for hair
103
removal in all cases. Preoperative hibiclens was not used routinely. Incisional closure was not
104
standardized and negative pressure incision management devices, such as PREVENA™, were
105
not utilized. The management of foot infection or gangrene present at the time of
106
revascularization was patient and surgeon specific. Most surgeons in our group prefer to perform
107
wound debridement and/or toe amputations as a separate procedure several days after the
108
revascularization. Standard follow up after lower extremity revascularization at our institution is
109
one and three months postoperatively, then six months thereafter. The social security death index
110
and hospital electronic medical records were used to record survival during the follow-up period.
111
Limb loss was defined as either below knee (BKA) or above knee amputation (AKA).
112
Summary statistics (e.g., mean and standard deviation, or median and interquartile range) were
113
calculated for continuous variables as well as frequencies and proportions for categorical
114
variables. The data for patient with grafts and patches was combined for this analysis. Subgroup
115
analysis was not performed due to the relatively small number of patients in our cohort.
116
Univariate and multivariable survival analysis methods were used to assess the association
117
between potential factors and risk of graft infection. Specifically, the Fine and Gray method for
118
competing-risks analysis was performed, whereby death was considered the competing event,
AC C
EP
TE D
M AN U
SC
RI PT
96
ACCEPTED MANUSCRIPT 6
and the event of interest is first instance of graft infection. Time-to-event was measured as the
120
number of days from surgery to graft infection (or mortality). Univariate subdistribution hazards
121
regression for competing risks was used for continuous variables. Factors that appeared to be
122
significantly associated with survival in the univariate analysis at alpha level 0.1 (including the
123
multiple comparisons tests), were considered for inclusion in the subdistribution hazards
124
regression multivariable model. Less strict p-value criteria was used during univariate analyses
125
to allow for identifying variables that may be confounded by another variable. Final model was
126
selected using the likelihood ratio test.
127
Cox regression extended for time-dependent covariates was performed to assess if presence of
128
graft infection influenced patient overall survival or limb loss where mortality was treated as
129
competing risk and graft infection status as time-varying covariate. A result was considered
130
significant at the P<.05 level of significance, unless otherwise stated. All analyses were
131
conducted using SAS version 9.4 (SAS Institute Inc., Cary, NC).
132
Results
133
A total of 272 patients underwent lower extremity reconstructions with 228 PTFE bypass grafts
134
and 44 Dacron patches. Of these, 58.8% were male with an average age of 72.6 ± 10.7 years.
135
64.3% of subjects were Caucasian. Median follow-up time was 668 days (IQR=588), ranging
136
from 0 to 1,363 days. Thirty-seven bypass grafts (16.2%) and six patches (13.6%) were found to
137
be infected. The overall rate of graft infection was 15.8% (43/272) with median time to
138
presentation of 43 days (IQE=85) after initial revascularization. The incidence of superficial
139
surgical site infection was 20.2% (55/272). Thirty five patients with graft infections had
140
concomitant superficial SSIs. The other 20 patients did not develop graft infections despite
141
documented signs of wound infection.
AC C
EP
TE D
M AN U
SC
RI PT
119
ACCEPTED MANUSCRIPT 7
Univariate competing risks analysis comparing patients with and without infection identified
143
female gender (P=.01), hypertension (P=.08), serum albumin level (P=.02), and redo surgery
144
(P=.01) as potential predictors of infection. No other comorbidities or patient characteristics were
145
associated with risk of infection, including race, age, insurance status, body mass index (BMI),
146
diabetes, heart disease, lung disease, renal insufficiency or active smoking (Table I). Presence of
147
acute limb ischemia and need for emergency surgery were not predictive of infection. The
148
presence of open wounds or gangrene on the operative limb was also not significantly associated
149
with the outcome (Table II). The type of revascularization performed or procedural
150
characteristics were not associated with infection either. The length of hospital stay prior to
151
procedure was similar in patients with and without infection. (Tables III and IV).
152
Average length of surgery was similar in both groups (238 min vs. 256 min, P=.26).
153
Perioperative antibiotics were administered appropriately in all the cases and Ancef was used
154
62.7% of the time. Chlorhexidine gluconate with isopropyl alcohol was used for skin
155
preparation in 74.0% of the cases. The rate and volume of intraoperative blood transfusions did
156
not correlate with risk of infection. In the group of patients without infections, 35.4% of them
157
received transfusions at an average of 2.2 units of blood per person. Similarly, in the graft
158
infection group, 37.2% of patients had transfusions at an average of 1.7 units per person.
159
Eighteen bypasses underwent revisions after the index procedure with similar rates in both
160
groups, 5.7% vs. 11.6% (P=.18).
161
The final multivariable model included redo surgery status, hypertension, gender, and albumin
162
level. The analysis suggests that the hazard for graft infection is 2.4 times greater among those
163
with a history of redo of surgery (95% CI: 1.3, 4.3, P=.01) and 2.1 times greater in females
164
compared to males (95% CI: 1.1, 3.8, P=.02) after adjusting for other factors in the model. A 1-
AC C
EP
TE D
M AN U
SC
RI PT
142
ACCEPTED MANUSCRIPT 8
unit increase in albumin level is associated with a 33.5% decrease in risk of infection (HR: .7,
166
95% CI: .5, .9, P=.03). Cumulative incidence was computed, holding albumin at the mean of
167
3.36 g/dl and comparing gender, hypertension status and history of redo surgery. The highest
168
incidence of infection was seen among female patients with hypertension undergoing redo
169
operation (Figure 1). For instance, estimated cumulative incidence of infection in this group was
170
19.4% at 30 days post-surgery (95% CI: 10.6, 35.6) and 39.9% at 1 year (95% CI: 26.8, 59.3).
171
Thirteen out of forty-three subjects (30.2%) suffered limb loss due to infection compared to the
172
8.7% (20/229) rate of amputation in patients without infections. The risk of limb loss was 8.1
173
times greater for patients with infections (95% CI for HR: 3.5, 18.5, P<.0001). Graft infection,
174
modeled as a time-dependent covariate, was also found to be a significant predictor of overall
175
survival. The rate of mortality during the follow period was 34.9% (15/43) in the infection group
176
vs. 13.1% (30/229) without infection. The risk of death was 6.4 times greater for a subject with
177
graft infection compared to one without infection (95% CI: 3.3, 12.3, P<.0001).
178
We employed several treatment options to manage bypass/patch infections, which was physician
179
and patient dependent (Table V). Thirty-two (76.2%) of the infected bypasses were patent on
180
presentation. Graft excision with or without reconstruction was performed in 53.5% of the cases,
181
while graft preservation was attempted in 39.5% of the cases. Graft preservation involved
182
aggressive wound debridement, followed by muscle flap coverage in 70.6% of the cases.
183
Vacuum assisted closure (VAC) devices were used in all the patients. Three patients were
184
severely sick due to infection and elected palliative care measures. The rate of limb loss was
185
significantly higher in the graft excision group as compared to the graft preservation group,
186
52.2% vs. 5.9% (P=.002). Mortality rates, however, were not affected by a chosen treatment
187
(30.4% vs 29.4%, P=.62). Staphylococcus was the most common bacteria, isolated in 51.2% of
AC C
EP
TE D
M AN U
SC
RI PT
165
ACCEPTED MANUSCRIPT 9
the cultures. Of those, 36.4% were methicillin-resistant Staphylococcus aureus (MRSA) (Table
189
VI). There was no correlation between culture results and patient outcomes.
190
Discussion
191
Surgical site infections (SSI) after lower extremity revascularization are relatively common and
192
well described in the literature, with reported incidence ranging from 4.8% to 15.6% [10-12].
193
Common factors associated with SSI have been identified in several recent studies, including the
194
data from Vascular Quality Initiative registry and American College of Surgeons National
195
Surgical Quality Improvement Program [10,11]. Transfusion rate, procedure time, skin
196
preparation, female gender, obesity, dialysis, and postoperative seroma are some of the common
197
factors associated with increased risk of SSI [10,11,13,14]. These superficial infections have
198
been shown to lead to deep infections involving prosthetic graft material used for
199
revascularization. As compared to SSI data, the literature on incidence, risk factors, and
200
outcomes of graft infections is less robust. No large prospective studies referring to these
201
variables are available at this time. Multiple previous studies did not distinguish wound
202
infections from graft infections, making it hard to interpret the results [8,9,15]. In our study,
203
twenty patients with wound infection did not develop graft infection, underscoring the fact that
204
these complications should be analyzed separately. Determining risks factors specific to graft
205
infection may allow us to reduce the rate of this troubling complication associated with high
206
morbidity and mortality rates.
207
During the three-year study period, we observed a 15.8% rate of prosthetic graft infections,
208
which is similar to previously published data [3-7]. We analyzed 44 patients with infections,
209
which is one of the largest cohort of patients reported in the literature to date. Female gender,
AC C
EP
TE D
M AN U
SC
RI PT
188
ACCEPTED MANUSCRIPT 10
210
redo surgery and low albumin were significant predictors of graft/patch infection in our
211
population. Our study suggests that females have more than a two-fold increase risk of infection as
213
compared to males. Similarly, Siracuse et al. reported 4.5 times higher rates of graft infection in
214
females [6]. This gender-related difference cannot be attributed to body weight alone, as BMI did
215
not correlate with infection in our study. Women have been shown to have greater lower
216
extremity fat distribution as compared to men [6,11,16], and may explain higher rates of
217
infection after lower extremity operations.
218
The association of redo surgery with prosthetic graft infection was also previously described in
219
the literature [3,6]. Often, these are technically challenging operations with prolonged dissection
220
time. Increased risk of infection may be due to poor vascularization and scarring encountered in
221
the re-operative field. As demonstrated by several studies, prolonged operative time correlates
222
with inferior healing rates and an increase in SSI [3,8,11,14].
223
It is not surprising that low albumin level in our study was associated with an increased risk of
224
infection. It is well established that hypoalbuminemia, as a marker of malnutrition, is associated
225
with a greater risk of adverse surgical outcomes. Results of the National VA Surgical Risk study
226
demonstrated that 1 g/dl decrease in albumin level was associated with more than a 2-fold
227
increase in mortality and morbidity. In the multivariate analyses, albumin level was a strong
228
predictor of major infection complications including systemic sepsis, pneumonia, and deep
229
wound infection [17]. An albumin level of less than 3.5 g/dl was found to be one of the fourteen
230
independent predictors of postoperative surgical site infection after general and vascular surgery,
231
based on results of the NSQIP patient safety in surgery study [18]. Several other factors that
AC C
EP
TE D
M AN U
SC
RI PT
212
ACCEPTED MANUSCRIPT 11
were not analyzed in our study have been shown to be more accurate predictors of nutritional
233
status, such as prealbumin, transferrin and retinol-binding protein levels, as well nutritional
234
assessment based on history and physical examination [19].
235
Prior studies demonstrated other potential risk factors for graft infection not observed in our
236
population. Early graft revision was reported as a predictor of prosthetic graft infections in
237
several studies. Brothers et al. described an eleven-fold increase in the risk of graft infection after
238
a bypass revision [4]. Kolakowski et al. reported an 11% graft infection rate in patients who had
239
a revision <30 days after the initial bypass operation versus 6.1% in those whose revision took
240
place >30 days later [5]. Based on our analysis, rates of bypass revision were not significantly
241
different in patient with graft infection as compared to those without infection. Diabetes, active
242
infection, limb amputation, and prolonged operative time, are other reported predictors for the
243
development of graft infection that were not observed in our study [3,4,6,8].
244
Management of infected prosthetic material continues to evolve. Complete excision of an
245
infected graft is still the preferred method of treatment; however, it places patients at a greater
246
risk for limb loss. Arterial reconstruction with extra-anatomic or in-situ graft replacement have
247
been described utilizing various conduits including biosynthetic prosthesis, cryopreserved human
248
allograft, prosthetic or a vein grafts [7,20-23]. In our study, more than two thirds of infected
249
bypasses were patent on presentation and graft preservation was attempted in 39.5% of these
250
cases. Mortality rates did not differ; however, rates of amputation were significantly lower in the
251
graft preservation group. The rates of repeated intervention and continued sepsis in the graft
252
preservation group was not recorded in our study, but likely to be higher, as reported by other
253
groups. Martens and colleagues described an increased incidence of subsequent operations for
254
continued sepsis in patient treated with incomplete removal of infected grafts; 82% vs. 13% [7].
AC C
EP
TE D
M AN U
SC
RI PT
232
ACCEPTED MANUSCRIPT 12
Several successful techniques of graft preservation have been described in the literature
256
including muscle flap coverage of the infected graft, and vacuum assisted closure. Morasch et al.
257
demonstrated an 85% rate of prosthetic graft salvage with the use of a gracilis muscle flap [24].
258
Similarly, coverage with a sartorius muscle flap was successful in 86% of cases reported by
259
Landry and colleagues [25]. Several other series described debridement and vacuum assisted
260
closure with rates of graft preservation ranging from 76% to 83% of patients [9,26]. Antibiotic-
261
loaded polymethyl methacrylate (PMMA) beads is another adjunct that can be used to improve
262
graft salvage rates [27]. Multiple series published in literature demonstrate high rates of
263
successful limb salvage utilizing these alternative treatment options. However, some
264
contradictory reports have been published. Taylor et al. described 78% rate of recurrent infection
265
after local debridement and muscle coverage was attempted [28]. At this time, it’s difficult to
266
draw definitive conclusions on outcomes of graft preservation techniques based on small,
267
retrospective series without long term follow up results.
268
Graft infections continue to be a major complication of revascularization procedures, particularly
269
those involving prosthetic material. In our population, these infections were associated with 30%
270
rate of limb loss and 35% mortality. These rates have not improved as compared to studies
271
published several decades ago [3,7,20]. Given the increased mortality and morbidity these
272
infections place on already vulnerable population, it is important to consider other
273
revascularization options and alternative conduits, when appropriate. Endovascular options, such
274
as lower extremity stents and angioplasty are worth considering. In the presence of diffuse
275
vascular disease, a bypass may be the only viable choice. The use of GSV as a bypass conduit
276
remains the gold standard, however, in the absence of an adequate GSV it may be worthwhile to
277
consider the use of other native veins, as autologous conduits demonstrate lower infection rates.
AC C
EP
TE D
M AN U
SC
RI PT
255
ACCEPTED MANUSCRIPT 13
278
Lloyd and colleagues demonstrated similar patency rates between ipsilateral GSV to opposite leg
279
veins, arm veins, lesser saphenous veins, superficial femoral veins, and spliced veins when used
280
as
281
This study is limited by its single center retrospective review design. The database is not entirely
282
comprehensive and certain parameters that may have potential impact on infection rates were not
283
evaluated, such as HbA1c, preoperative medication, WiFi classification of tissue loss, nutritional
284
parameters, etc. Similarly, we cannot account for some important procedural details such as
285
preparation and draping of the surgical field, management of infected or gangrenous tissues, and
286
skin closure techniques. There are several potential confounders outside our study design that we
287
cannot account for, such as technical skill of the surgeons, intraoperative management and
288
anesthesia, as well as postoperative incisional care.
289
Conclusion
290
Female gender, redo surgery and malnutrition are associated with increased risk of prosthetic
291
graft infections leading to high rates of limb loss and mortality. Endovascular interventions and
292
bypasses with vein conduits should be considered in these patients.
conduits
[2].
AC C
EP
TE D
M AN U
SC
RI PT
alternative
ACCEPTED MANUSCRIPT 1
References 1. Reed JF, Eid S, Edris B, et al. Prevalence of peripheral artery disease varies significantly
Rehabil 2009;16:377-81. doi:10.1097/hjr.0b013e32832955e2.
RI PT
depending upon the method of calculating ankle brachial index. Eur J Cardiovasc Prev
2. Taylor LM, Edwards JM, Brant E, et al. Autogenous reversed vein bypass for lower
1987;153:505–10. doi:10.1016/0002-9610(87)90803-8.
SC
extremity ischemia in patients with absent or inadequate greater saphenous vein. Am J Surg
M AN U
3. Edwards WH Jr, Martin RS, Jenkins JM, et al. Primary graft infections. J Vasc Surg 1987;6:235-9. doi:10.1016/0741-5214(87)90034-6.
4. Brothers T, Robison J, Elliott B. Predictors of Prosthetic Graft Infection after Infrainguinal Bypass. J Am Coll Surg 2009;208:557-61. doi:10.1016/j.jamcollsurg.2009.01.001. 5. Kolakowski S Jr, Dougherty MJ, Calligro KD. Does the timing of reoperation influence the
TE D
risk of graft infections? J Vac Surg 2007;45:60-4. doi:10.1016/j.jvs.2006.09.007. 6. Siracuse JJ, Nandiva P, Giles KA, et al. Ten-year experience with prosthetic graft infections involving the femoral artery. J Vasc Surg 2013;57:700-5. doi:10.1016/j.jvs.2012.09.049.
EP
7. Mertens RA, O’Hara PJ, Hertzer NR, et al. Surgical management of infrainguinal arterial
AC C
prosthetic graft infections: review of a thirty-five-year experience. J Vasc Surg 1995;21:78290.
8. Chang JK, Calligaro KD, Ryan S, et al. Risk factors associated with infection of lower extremity revascularization: analysis of 365 procedures performed at a teaching hospital. Ann Vasc Surg 2003;17:91-6. doi:10.1007/s10016-001-0337-8.
ACCEPTED MANUSCRIPT 2
9. Dosluoglu HH, Loghmanee C, Lall P, et al. Management of early (<30 day) vascular groin infection using vacuum-assisted closure alone without muscle flap coverage in consecutive patient series. J Vasc Surg 2010;51:1160-6. doi:10.1016/j.jvs.2009.11.053.
RI PT
10. Kalish JA, Farber A, Homa K, et al. Factors associated with surgical site infection after lower extremity bypass in the Society for Vascular Surgery (SVS) Vascular Quality Initiative (VQI). J Vasc Surg 2014;60:1238-46. doi:10.1016/j.jvs.2014.05.012.
lower
extremity
revascularization.
J
Vasc
Surg
2011;54:433–9.
M AN U
doi:10.1016/j.jvs.2011.01.034.
SC
11. Greenblatt DY, Rajamanickam V, Mell MW. Predictors of surgical site infection after open
12. Adam DJ, Beard JD, Cleveland T, et al. Bypass versus angioplasty in severe ischemia of the leg (BASIL): multicenter, randomized controlled trial. Lancet 2005;366:1925–34. doi:10.1016/j.jvs.2010.01.077.
procedures
TE D
13. Kuy S, Dua A, Desai S, et al. Surgical site infections after lower extremity revascularization involving
groin
incisions.
Ann
Vasc
Surg
2014;28:53-8.
doi:10.1016/j.avsg.2013.08.002.
EP
14. Davis FM, Sutzko DC, Grey SF, et al. Predictors of surgical site infection after open lower extremity revascularization. J Vasc Surg 2017;65:1769-78. doi:10.1016/j.jvs.2016.11.053.
AC C
15. Jensen LJ, Kimose HH. Prosthetic graft infections: a review of 720 arterial prosthetic reconstructions. Thorac Cardiovasc Surg 1985;33:391-8. doi:10.1055/s-2007-1014176. 16. Nguyen LL, Brahmanandam S, Bandyk DF, et al. Female gender and oral anticoagulants are associated with wound complications in lower extremity vein bypass: an analysis of 1404 operations
for
critical
doi:10.1016/j.jvs.2007.07.053.
limb
ischemia.
J
Vasc
Surg
2007;46:1191–7.
ACCEPTED MANUSCRIPT 3
17. Gibbs J, Cul W, Henderson W, et al. Preoperative Serum Albumin Level as a Predictor of Operative Mortality and Morbidity. Arch Surg 1999;134:36-42. 18. Neumayer L, Hosokawa P, Itani K, et al. Multivariable Predictors of Postoperative Surgical
RI PT
Site Infection after General and Vascular Surgery: Results from the Patient Safety in Surgery Study. J Am Coll Surg 2007;204:1178-87. doi:10.1016/j.jamcollsurg.2007.03.022.
19. Bharadwaj S, Ginoya S, Tandon P, et al. Malnutrition: laboratory markers vs. nutritional
SC
assessment. Gastroenterology Report 2016;4:272-80. doi:10.1093/gastro/gow013.
infection. AM Surg 1998;645:39-45.
M AN U
20. Henke PK, Bergamini TM, Rose SM, et al. Current options in prosthetic vascular graft
21. Topel I, Betz T, Uhl C, et al. Use of biosynthetic prosthesis to replace infected infrainguinal prosthetic grafts-- first results. VASA 2012;41:212-20. doi:10.1024/0301-1526/a000188. 22. Wiltberg G, Matia L, Schmelzle M, et al. Mid- and long-term results after replacement of
TE D
infected peripheral vascular prosthetic grafts with biosynthetic collagen prosthesis. J Cardiovasc Surg 2014;55:693-8.
23. Brown KE, Heyer K, Rodriguez H, et al. Arterial reconstruction with cryopreserved human
EP
allograft in the setting of infection: a single-center experience with midterm follow-up. J Vasc Surg 2009;49:660-6. doi:10.1016/j.jvs.2008.10.026.
AC C
24. Morasch MD, Sam AD, Kibbe MR, et al. Early results with use of gracilis muscle flap coverage of infected groin wound after vascular surgery. J Vasc Surg 2004;39:1277-83. doi:10.1016/j.jvs.2004.02.011. 25. Landy GJ, Carlson JR, Liem TK, et al. The sartorius muscle flap: an important adjunct for complicated femoral wounds involving vascular grafts. Am J Surg 2009;197:655-9. doi:10.1016/j.amjsurg.2008.12.020.
ACCEPTED MANUSCRIPT 4
26. Acosta S, Monsen C. Outcome after VAC therapy for infected bypass grafts in the lower limb. Eur J Vasc Endovasc Surg 2012;44:294-9. doi:10.1016/j.ejvs.2012.06.005.
RI PT
27. Stone PA, Mousa AY, Hass SM, et al. Antibiotic-loaded polymethylmethacrylate beads for treatment of extracavitary vascular surgical site infections. J Vasc Surg 2012;55:1706-11. doi:10.1016/j.jvs.2011.12.037.
SC
28. Taylor SM, Weatherford DA., Langan EM, et al. Outcomes in the management of vascular prosthetic graft infection confined to the groin: a reappraisal. Ann Vasc Surg 1996; 10:117-
AC C
EP
TE D
M AN U
22. doi:10.1007/BF02000754.
ACCEPTED MANUSCRIPT Characteristics
Patients without infection N= 229 n (%)
Patients with infection N = 43 n (%)
SHR with 95% CI, P-value*
72.2 ±10.9
74.6 ± 9.07
1.0 (.09-1.0), .12
87 (38.0)
25 (58.1)
2.1 (1.2-3.9), .01
Caucasian
148 (64.6)
29 (67.4)
n/a
African American
46 (20.1)
10 (23.3)
1.1 (.5-2.3), .75
Asian
17 (7.4)
1 (2.3)
.3 (.05-2.2), .25
Hispanic
5 (2.2)
Age (years), mean ±SD Sex (% female)
SC
RI PT
Race
3 (7)
Medicare
170 (74.2)
Medicaid
30 (13.1)
Private
29 (12.7)
M AN U
Insurance
.8 (.2-3.3), .73
36 (83.7)
1.5 (.5-4.2), .46
3 (7)
.7 (.2-3.1), .65
4 (9.3)
n/a
26 (60.5)
1.6 (.9-2.9), .14
194 (84.7)
41 (95.4)
3.5 (.8-14.2), .08
96 (41.9)
17 (39.5)
.9 (.5-1.7), .8
28 (12.2)
9 (20.9)
1.8 (.8-3.7), .13
45 (19.7)
8 (18.6)
.4 (.1-1.6), .2
42 (18.3)
9 (20.9)
1.1 (.5-2.2), .8
BMI (kg/m2), mean ±SD
26.9 ±5.6
28.2 ± 5.2
1.0 (1.0-1.1), .13
Albumin (g/dl), mean ±SD
3.4 ±0.7
3.1 ± 0.7
.64 (.4-.9), .02
111 (48.5)
Hypertension Coronary artery disease
EP
COPD
TE D
Diabetes Mellitus
Renal Disease
AC C
Current Smoker
*The Fine and Gray proportional subdistribution hazard model was used with all-cause death as a competing risk.
ACCEPTED MANUSCRIPT
Patients with infection N = 43 n (%)
SHR with 95% CI, P-value*
Claudication
31 (13.5)
4 (9.3)
.9 (.2-3.4), .84
Rest pain
39 (17.0)
8 (18.6)
1.3 (.4-3.9), .68
Open Wound
55 (24.0)
13 (30.2)
1.4 (1.4-.5), .5
Gangrene
47 (20.5)
10 (23.3)
1.3 (1.3-.4), .61
23 (10)
3 (7)
.7 (.2-2.1), .5
Aneurysms Acute limb ischemia
RI PT
Patients without infection N = 229 n (%)
SC
Indication for Surgery
34 (14.9)
5 (11.6)
.8 (.3-2.1), .62
M AN U
*The Fine and Gray proportional subdistribution hazard model was used with all-cause death as
AC C
EP
TE D
a competing risk.
ACCEPTED MANUSCRIPT
Type of Surgery
Patients without infection N = 229 n (%)
Patients with infection N = 43 n (%)
Aorta-Femoral
7 (3.1)
0
Ilio-Femoral
19 (8.3)
2 (4.7)
Axillary-Femoral
11 (4.8)
5 (11.6)
SHR with 95% CI, P-value*
Femoral-Femoral
37 (16.2)
4 (9.3)
Femoral-Popliteal
79 (34.5)
14 (32.6)
RI PT
.3 (.1-1.2), .09
Femoral-Tibial
38 (16.6)
12 (27.9)
n/a
SC
.6 (.3-1.5), .32
.6 (.3-1.3), .19
M AN U
Femoral 38 (16.6) 6 (14) .6 (.2-1.5), .26 Endarterectomy *The Fine and Gray proportional subdistribution hazard model was used with all-cause death as
AC C
EP
TE D
a competing risk.
Patients without infection N = 229 n (%)
Patients with infection N = 43 n (%)
SHR with 95% CI, P-value*
142 (62.0)
26 (60.5)
1.0 (.5-1.9), .95
Vancomycin
34(14.8)
6 (14.0)
n/a
Other
53 (23.2)
11 (25.5)
n/a
Chlorhexidine gluconate
170 (74.2)
29 (67.4)
n/a
Betadine
59 (25.8)
SC
ACCEPTED MANUSCRIPT Characteristics
14 (32.6)
1.2 (.6-2.3), .6
Procedure Time (min), mean ±SD
238 ± 113
256 ± 130
1.1 (.9-1.2), .26
Rate of blood transfusions
81 (35.4)
16 (37.2)
.9 (.8-1.1), .5
Time from admission to surgery(d), median(range)
3.0 (0-8)
1.0 (0-7)
1.0 (1.0-1.1), .19
Redo surgery
49 (21.4)
17 (39.5)
2.2 (1.3-4.3), .01
13 (5.7)
5 (11.6)
1.8 (.8-4.1), .18
Ancef
TE D
Bypass Revision
M AN U
Skin preparation
RI PT
Perioperative Antibiotics
AC C
EP
*The Fine and Gray proportional subdistribution hazard model was used with all-cause death as a competing risk.
ACCEPTED MANUSCRIPT
Limb Loss n(%)
Mortality n(%)
Graft removal ± revascularization
23 (53.5)
12 (52.2)
7 (30.4)
No revascularization
15 (65.2)
12 (80)
5 (33.3)
In-situ bypass with vein
4 (17.4)
0
In-situ bypass with homograft
2 (8.7)
0
Extra-anatomic bypass
2 (8.7)
0
Graft preservation
17 (39.5)
RI PT
n(%)
SC
Treatment (N=43)
2 (50.0) 0 0
1 (5.9)
5 (29.4)
1 (5.9)
3 (25.0)
12 (70.6)
Debridement + VAC
5 (29.4)
0
2 (40.0)
3 (7)
0
3 (100)
M AN U
Debridement + muscle flap+ VAC
AC C
EP
TE D
Palliative care
ACCEPTED MANUSCRIPT
All Staphylococcus species
N=43 (%) 51.2
Methicillin- Resistant Staphylococcus aureus
18.6
Enterococcus
20.9
RI PT
Bacteria Isolated
Pseudomonas aeruginosa
18.6
Proteus
16.3 11.6
SC
E.coli Enterobacter
4.7
M AN U
Yeast Streptococcus Klebsiella pneumoniae Morganella morganii
Corynebacterium Polymicrobial
AC C
EP
No growth
TE D
Acenetobacter
4.7 2.3 2.3 2.3 2.3 2.3
34.9 4.7
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT