Transcutaneous Oxygen (TcPO2) Estimates Probability of Healing in the Ischemic Extremity

JOURNAL OF SURGICAL RESEARCH

60, 365–369 (1996)

Article No. 0059

Transcutaneous Oxygen (TcPO2) Estimates Probability of Healing in the Ischemic Extremity FRANK T. PADBERG, JR., M.D.,1 THOMAS L. BACK, M.S., R.V.T., PETER N. THOMPSON, M.D., AND ROBERT W. HOBSON II, M.D. East Orange VAMC and New Jersey Medical School, UMDNJ, Newark, New Jersey 07103 Presented to the Annual Symposium of the Association of Veterans Administration Surgeons, Pittsburgh, Pennsylvania, May 1–3, 1995

The severity of arterial ischemia is a major variable affecting healing of extremity wounds. By relating risk of failure to severity of ischemia, the probability of wound healing may be stratified along with assessment of general medical risks. Transcutaneous oxygen tension (TcPO2) arterial segmental pressure (ASP), and arterial segmental indices (ASI) were obtained in 204 ischemic lower extremity sites; 63% of the sites were in patients with diabetes mellitus (DM), 11% in patients with chronic renal failure (CRF), and 37% in patients with neither DM nor CRF (ND). Wounds included 126 amputations and 78 gangrenous ulcerations of the foot or toes. Healing (n Å 112) was defined as complete wound closure, without regard to the time required. Failure (n Å 92) was defined by the requirement for either arterial reconstruction (n Å 45) or proximal amputation (n Å 47). Stepwise multiple regression analysis was used to assess the relative contribution of each measurement and to predict the probability of healing; TcPO2 was superior to ASP and ASI in all categories. TcPO2 was the only test meeting the P õ 0.05 entry criteria modeled by the regression. An accuracy of 83% was achieved. However, when each test was evaluated by univariate analysis, ASP and ASI did meet the criteria for the ND group. However, the accuracy was 68 and 72%, respectively. Predictive accuracy of TcPO2 was unaffected by DM or CRF. ASP and ASI were satisfactory in the ND group, although of slightly reduced accuracy. ASP and ASI were misleading and inaccurate in DM and CRF. Thus, of the noninvasive tests, TcPO2 alone is sufficient for objective risk stratification of arterial ischemia in the lower extremity. q 1996 Academic Press, Inc.

INTRODUCTION

Tissue growth and metabolism are retarded in arterial insufficiency and wound healing will fail completely with severe ischemia. Arterial reconstruction, however, may alter the balance between limb salvage 1 To whom correspondence and reprint requests should be addressed at VA Medical Center, 385 Tremont Ave., East Orange, NJ 07019.

and amputation. The highest risk wounds are often found in the highest risk patients, those with diabetes mellitus and chronic renal insufficiency. Objective determination of the risk of limb loss would provide valuable adjunctive information for clinical assessment [1 – 5]. The morbidity and mortality of arterial reconstruction is most frequently related to the severity of coronary arterial disease. Myocardial risk stratification offers a model for management based on risk assessment [6, 7]. Similar attempts to categorize anesthetic risks are widely used but less well substantiated [8]. Individual patients may respond differently when presented with estimates of surgical risk. For example, when the risks of limb loss are high, greater medical risk may be acceptable. As with general medical risk stratification, the severity of arterial ischemia can also be integrated into a comprehensive management plan. Methods used to assess limb ischemia have included Doppler-derived arterial segmental pressures (ASP), as well as skin blood flow with laser Doppler, toe pressures, pulse volume recordings (PVR), Xenon washout, and transcutaneous oxygen tension (TcPO2) [1–3, 5, 9– 15]. TcPO2 measurements correlate directly with progressive ischemia as determined experimentally [10, 16]. Doppler-derived arterial segmental pressures are generally performed with reproducible results which are widely accepted [5, 15, 17]. While digital arterial pressures have been successfully employed, prior amputations or trophic skin changes of digits may limit uniform application of this method [12]. PVR recordings are also employed but they are subjective and not quantitative [5, 11, 15]. Varied criteria for the prediction of wound healing have been recommended for TcPO2 ranging from 10 mm Hg [1, 2, 13], to 20 mm Hg [4, 14], to 40 mm Hg [15]. Although a clear relationship exists between increasing severity of ischemia and diminished TcPO2 , a critical point for tissue viability has not been agreed upon. Arterial segmental pressures and indices are useful measures of arterial insufficiency. However, a reliable index or pressure criteria for the determination of tissue viability has varied widely [5]. Minimal perfusion 0022-4804/96 $12.00 Copyright q 1996 by Academic Press, Inc. All rights of reproduction in any form reserved.

365

AID

JSR 4705

/

m4725$$221

01-19-96 20:01:21

srga

AP: Surg Res

366

JOURNAL OF SURGICAL RESEARCH: VOL. 60, NO. 2, FEBRUARY 1, 1996

TABLE 1 Wound Type and Outcome Wounds

Healed

All wounds Amputations AKA BKA TMA Toe Ulcerations Toe Foot Coexisting medical disease Diabetes (DM) Renal failure (CRF) Neither (ND) Diabetes without renal failure (DMA)

Failed (F)

F Å Amp

92 (45%) 32 3 2 12 15 60 46 14

47 16 3 2 7 4 31 25 6

45 16 0 0 5 11 29 21 8

204 126 (62%) 37 15 26 48 78 (38%) 59 19

70 11 42

59 11 33

29 10 18

30 1 15

129 (63%) 22 (11%) 75 (37%)

59

48

19

29

107 (52%)

112 (55%) 94 34 13 14 33 18 13 5

F Å Rec

Total

Comparison of healed and failed subgroups TcPO2 (mm Hg) Segment pressure (ASP, mm Hg) Segment index (ASI)

39 { 2

9{1

10 { 2

8{2

P õ 0.0001

131 { 7.0 0.91 { 0.05

102 { 9.5 0.70 { 0.06

107 { 15 0.73 { 0.1

97 { 11 0.67 { 0.8

P õ 0.0165 P õ 0.0102

Note. Mean values between healed and failed groups were significantly different for each noninvasive measurement; P values for this comparison are given in the last column. Amp, amputation; Rec, arterial reconstruction. Percentages cited are of the whole data set (n Å 204).

pressures to achieve healing have been reported as 30 mm Hg (index 0.34) [2] to 70 mm Hg (index 0.35); a higher threshold (0.45) was recommended for diabetic patients [13]. Others were unable to define a criteria because of overlap in the data [9]. We propose the use of a probability approach in which the likelihood of wound healing is estimated by the laboratory measurement of TcPO2 . By implication, this approach also categorizes a level of risk for wound failure. This prediction can then be integrated with the clinical assessment and a general management plan formulated. METHODS Between June 1989 and December 1993, 204 wounds in critically ischemic lower extremities were assessed by transcutaneous oxygen tension (TcPO2 ), arterial segmental pressure (ASP), and arterial segmental indices (ASI). Heel ulcers, limbs with uncontrolled foot infections, or those with an obvious neuropathic etiology were excluded since these factors may preclude healing at the local site. Wounds evaluated included 126 amputations distributed as 37 AKA, 15 BKA, 26 TMA, and 48 toes and 78 gangrenous ulcerations of the foot or toes. Diabetes mellitus (DM) was present as a coexisting condition for 129 (63%) of the sites. Dialysis dependent chronic renal failure (CRF) was present as a coexisting condition for 22 (11%) of the sites; 25 of the sites were associated with creatinine levels ú3.0 mg/dl, and 38 were associated with a creatinine ú2.0 mg/dl, of which all but 2 were also associated with DM. Sites associated with DM but without CRF (DMA, n Å 107) were also assessed to eliminate overlap between groups. Neither disease (ND) was present as a coexisting condition for 75 of the sites. These data are summarized in Table 1. Healing (n Å 112) was defined as complete wound closure, with epithelialization of the wound surface or a healed suture line. Failure

AID

JSR 4705

/

m4725$$222

01-19-96 20:01:21

to heal (n Å 92) was defined by the requirement for either arterial reconstruction (n Å 45) or proximal amputation (n Å 47). Definitive end points were obtained for all 204 sites. An additional eight sites had indeterminate outcomes resulting from loss to follow-up (3) or mortality (5) prior to achieving an end point; since outcome of these eight wounds will never be known, these patients were not included in the statistical evaluation. The 204 sites were distributed among lower extremities of 137 patients, 64% (87) of whom had diabetes and 9% (12) of whom had dialysis-dependent chronic renal failure. A healed end point was not entered until 1 month following amputation to assure that the wound remained healed following suture removal. Likewise, ulcerations or open amputation sites were not entered until closed by epithelial ingrowth or split thickness skin grafts. Wounds which failed to heal were often obvious within 2 to 3 weeks. Accordingly, the mean interval to an end point was greater for healed wounds at 130 days (range 30–501 days) than that for wounds which failed to heal (mean of 30 days, range 1–275 days). While the results of noninvasive determinations were known to the surgeons caring for these patients, the decision regarding amputation, revascularization, or observation was based on clinical assessment of the wound and the medical status of the individual patient. TcPO2 determinations were performed with an oxymonitor (SM 361, Litton Datamedix, Sharon, MA). After cleansing the measurement site with alcohol, the probe was applied, and the skin heated to 457C. When a stable steady state was achieved, a value expressed in millimeters of mercury (mm Hg) was transcribed. This value is a measurement of the partial pressure of oxygen on the skin surface. Patients were studied in the supine position breathing room air. Measurement sites and representative traces were illustrated previously [1]. Doppler-derived ASPs were determined using a narrow 7-cm-wide cuff on the thigh and a standard adult cuff on the calf and ankle; the higher of the two ankle systolic pressures was used by convention. This methodology was based on prior work from our noninvasive laboratory [1, 2, 17]. Cuff pressures were indexed against the brachial systolic pressure to obtain an ASI appropriate to the site. Stepwise multiple logistic regression analysis was conducted to determine the best noninvasive testing modality between ASP, ASI,

srga

AP: Surg Res

367

PADBERG ET AL.: TcPO2 AND PROBABILITY OF HEALING

FIG. 1. Estimated probability of healing with TcPO2 , in diabetes mellitus (DM), chronic renal failure (CRF), and neither DM nor CRF (ND).

and TcPO2 for each group and subgroup. The criteria for acceptance of a test into the stepwise model was P õ 0.05. Univariate logistic regression was then run for each test. Values obtained were then plotted against outcome as healed or failed; an equation and representative plot were obtained using SAS version 6.09 software (SAS Institute, Cary, NC). Each testing modality was assessed separately for the whole population, and in groups consisting of DM, CRF, and ND. Probability curves were compared using multiple logistic regression using three independent variables. Mild renal failure was defined as serum creatinine ú2.0 and ú3.0 mg/dl; these subgroups were also evaluated. The equation

PÅ

1 1 100 1 / e0(a / bx)

estimated probability (in %) of healing (P) at a measured value of x; e represents the natural logarithm (approximately 2.718); a and b are constants. Results are expressed in terms of predictive accuracy and probability of healing.

FIG. 2. Estimated probability of healing with ASI in diabetes mellitus (DM), chronic renal failure (CRF), and neither DM nor CRF (ND). The regression plot for CRF ends at 0.36 since there were no observations below this level. Lack of correlation with healing in CRF is emphasized by the implication that a lower ASI would improve probability of wound healing.

RESULTS

TcPO2 was the best noninvasive examination for determining the likelihood of healing or failure to heal the wounds. The probability of healing for TcPO2 is presented as Fig. 1. The TcPO2 curves are almost superimposed; the constants and P values are given in Table 2. There was no significant uniform shift between the curves for the groups of patients with DM, DMA, CRF, or ND; DM vs ND, P Å 0.61; CRF vs ND, P Å 0.61; DMA vs CRF, P Å 0.79; and DMA vs ND, P Å 0.70. In contrast, the predictive capability of ASP and ASI was different between DM, CRF, and ND (Fig. 2). For

TABLE 2 Predictive accuracy and Probability Constants Predictive Accuracy (%)

No. sites TcPO2 DM 129 CRF 22 ND 75 ASP DM 129 CRF 22 ND 75 ASI DM 129 CRF 22 ND 75 TcPO2 only reported Crú2 38 All 204

Equation constants

Accuracy

Sensitivity

Specificity

a

b

P value

81 77 84

81 73 86

81 82 82

01.8995 01.7246 02.2522

0.0952 0.0710 0.1159

0.0001 0.0180 0.0001

61 0 68

84 0 71

32 0 64

00.2127 0.2428 02.1703

0.00268 00.00114 0.0358

0.1968 0.7855 0.0001

58 50 72

81 46 74

31 55 70

00.1985 0.7918 02.0895

0.3804 00.5984 4.7758

0.2077 0.3746 0.0001

84 83

85 84

83 82

01.9010 02.0188

0.0919 0.1020

0.0011 0.0001

Note. Predictive accuracy of TcPO2 , ASP, and ASI is given from the 50% level. Constants a and b are terms employed in the calculation of probability and are specific to each group.

AID

JSR 4705

/

m4725$$222

01-19-96 20:01:21

srga

AP: Surg Res

368

JOURNAL OF SURGICAL RESEARCH: VOL. 60, NO. 2, FEBRUARY 1, 1996

FIG. 3. Estimated probability of healing with TcPO2 in all 204 sites. Upper and lower curves delineate 95% confidence limits.

DM and CRF, pressures or indices failed to discriminate between healing and failing. For patients with mild chronic renal failure, pressures or indices also failed to discriminate between healing and failing. For patients without diabetes or renal failure (ND), arterial segmental pressures or indices did reflect the probability of healing. However, these predictions were less accurate than that of TcPO2 (Table 2). Since there was little difference in the predictive accuracy, only ASI is presented graphically for comparison to probability of healing with TcPO2 ; the nearly horizontal curves for DM and CRF demonstrate the nondiagnostic value of indices in these patients (Fig. 2). For patients with mild renal insufficiency, TcPO2 was the only examination meeting the criteria for selection in stepwise multiple logistic regression. When evaluated with univariate logistic regression, pressurebased tests produced an unacceptable level of accuracy (Table 2). TcPO2 alone was as accurate as any combination of noninvasive examinations whether investigated by stepwise multiple regression or nonstepwise multiple regression. The probability of healing for all wounds using TcPO2 is presented with the 95% confidence limits as Fig. 3. DISCUSSION

Oxygen is an essential substrate for wound healing. However, the critical hypoxic level which completely shuts off regeneration or angiogenesis is unknown. Wound healing and tensile strength develop faster in a richer oxygen environment and more slowly when oxygen is limited in availability [18]. The TcPO2 is reduced in severe arterial insufficiency, but the decrease is not linear, dropping precipitously when arterial flow falls below 20% of normal [16]. One investigator [19] reported the association of a rising partial pressure of oxygen during the progressive revascularization of split thickness cutaneous autografts. Similarly, revascularization of the critically ischemic limb is rewarded by the dramatic transformation of a pale, necrotic ulcer into a vibrant bed of regenerating capillaries. However,

AID

JSR 4705

/

m4725$$222

01-19-96 20:01:21

this same group of elderly patients are also at high risk for myocardial and cerebrovascular events. Therefore, objectively estimating the probability of healing by TcPO2 becomes useful in clinical management. Tissue oxygen determinations using a heated, noninvasive probe are reliable and reproducible [20]. Its ready adaptability and application at different sites adds to its appeal. The test is not difficult to master, but may require several minutes to achieve adequate stabilization. Despite relative uniformity in instrumentation, multiple criteria have been proposed by authors studying TcPO2 . This variability may be explained by an emphasis on identification of a single value to best separate healed wounds from those which failed to heal. Prevalence of wounds which failed to heal in some studies has been less than 15% [13, 15]; another evaluated 24 successful amputations but curiously reported no failures [14]. Some have altered patient position to dependency or increased inspired oxygen during the examination to enhance the reliability of the TcPO2 [13]. Age was a significant variable in some [15], but not in others [13]. Time to healing was factored into some studies; if a wound was not healed within 1 month [2, 9] or if it had a secondary breakdown, it was classed as a failure to heal. All of these factors may contribute to variance in reported criteria. Neither ASP nor ASI is as good as TcPO2 ; but their accuracy is reasonable in the absence of DM or CRF (ND). Multiple cuff sites are essential in the use of Doppler segmental pressure assessment as described here. Pseudohypertension is the predominant cause of inaccurate predictions in patients with diabetes and chronic renal failure. Alternate testing modalities are recommended for adjunctive noninvasive assessment in these subgroups. Skin blood flow is another useful parameter which is used to predict outcome of amputations [1–3, 14]. Similar methodology estimating the probability of healing has been described using the laser Doppler with a heated probe (LDHP) [3]. These data confirm that a TcPO2 of 0 mm Hg is unable to provide an absolute prediction of failure to heal; i.e., a few wounds will still heal despite a TcPO2 of 0 mm Hg [1, 13, 15]. Thus, LDHP may discriminate failure more accurately in limbs with severe ischemia. While not widely adopted in the United States the available instrumentation and its application is not difficult to master. Toe pressures or indices, while useful, are limited by absence of digits on which to place a cuff. Of note, a probability estimation using digital pressure determination as one of several other variables has also been described [12]. Xenon-133 washout has been difficult to use clinically and its reproducibility has been seriously questioned [9]. PVR, although widely used, suffers from the inability to quantitate its waveform [5, 15]. Today’s sophisticated patient understands and appreciates the concept of risk management in relation to likelihood of wound healing. Goldman [6] described a multivariate analysis which identified patients at high risk for adverse perioperative events. Eagle concen-

srga

AP: Surg Res

PADBERG ET AL.: TcPO2 AND PROBABILITY OF HEALING

trated on cardiac factors and identified five clinical variables and two variables derived from dipyridamole thallium stress tests [7]. Anesthesiologists have attempted to predict risk based upon a scheme using five generalized classifications and attempted to relate this to perioperative mortality [8]. These attempts to define general medical risk imply that different actions will result directly from this stratification. In conclusion, of the noninvasive tests, TcPO2 alone is sufficient for objective risk stratification of ischemic extremity wounds. ASP or ASI are satisfactory for use in patients without DM or CRF; however, these investigations can be misleading or inaccurate in patients with DM or CRF. Using these data, critically ischemic limbs may be stratified by TcPO2 for the probability of healing (or risk of failure) using the curve presented (Fig. 3). This information may then be integrated with the risk of surgical intervention from an individualized standpoint. ACKNOWLEDGMENT The assistance of Mr. Martin Feuerman from the New Jersey Medical School Academic Computer Center in the statistical analysis of the data is greatly appreciated.

REFERENCES 1. Padberg, F. T., Back, T. L., Hart, L. C., and Franco, C. D. Comparison of heated-probe laser Doppler and transcutaneous oxygen measurements for predicting outcome of ischemic wounds. J. Cardiovasc. Surg. 33: 715, 1992. 2. Karanfilian, R. G., Lynch, T. G., Zirul, V. T., Padberg, F. T., Jr., Jamil, Z., and Hobson, R. W., II. Value of laser Doppler velocimetry and transcutaneous oxygen tension determination in predicting healing of ischemic forefoot ulcerations and amputations in diabetics and non-diabetics. J. Vasc. Surg. 4: 511, 1986. 3. Back, T. L., Padberg, F. T., Thompson, P. N., and Hobson, R. W., II. Probability of successful wound outcome determined by laser Doppler measurements with a heated probe (LDHP). J. Vasc. Tech. 18: 67, 1994. 4. Wyss, C. R., Harrington, R. M., Burgess, E. M., and Matsen, F. A., III. Transcutaneous oxygen tension as a predictor of success after amputation. J. Bone Jt. Surg. 70A: 203, 1988. 5. Yao, J. S. T. Choice of amputation level: Editorial. J. Vasc. Surg. 8: 544, 1988.

AID

JSR 4705

/

m4725$$222

01-19-96 20:01:21

369

6. Goldman, L. Cardiac risks and complications of non-cardiac surgery. Ann. Surg. 198: 780, 1983. 7. Eagle, K. P., Coley, C. M., Newell, J. B., et al. Combining clinical and Thallium data optimizes preoperative assessment of cardiac risk before major vascular surgery. Ann. Int. Med. 110: 859, 1989. 8. Goldstein, A., and Keatas, A. S. The risk of anesthesia. Anesthesiology 38: 130, 1970. 9. Malone, J. M., Anderson, G. J., Lalka, S. G., Hagaman, R. M., Henry, R., McIntyre, K. E., and Bernhard, V. M. Prospective comparison of non-invasive techniques for amputation level selection. Am. J. Surg. 154: 179, 1987. 10. Matsen, F. A., III, Wyss, C. R., Pedegana, L. R., Krugmire, R. B., Simmons, C. W., King, R. V., and Burgess, E. M. Transcutaneous oxygen measurement in peripheral vascular disease. Surg. Gynecol. Obstetr. 150: 525, 1980. 11. Rivers, S. P., Veith, F. J., Ascer, E., and Gupta, S. Successful conservative treatment of severe limb-threatening ischemia: Value of non-sympathectomy. Surgery 99: 759, 1986. 12. Vitti, M. J., Robinson, D. V., Hauer-Jensen, M., Thompson, B. W., Tanval, T. W., Barone, G., Barnes, R. W., and Eidt, J. Wound healing in forefoot amputations and predictive value of toe pressure. Ann. Vasc. Surg. 8: 99, 1994. 13. Oishi, C. S., Fronek, A., and Golbranson, F. L. Role of noninvasive vascular studies in determining levels of amputation. J. Bone Jt. Surg. 70A: 1520, 1988. 14. Lantsberg, L., and Goldman, M. Laser Doppler flowmetry, transcutaneous oxygen tension measurements and Doppler pressure compared in patients undergoing amputations. Eur. J. Vasc. Surg. 5: 195, 1991. 15. Cina, C., Katsamouris, A., Megerman, J., Brewster, D. C., Strayhorn, E. C., Robison, J. G., and Abbott, W. M. Utility of transcutaneous oxygen tension measurement in peripheral arterial occlusive disease. J. Vasc. Surg. 1: 362, 1984. 16. Moosa, H. H., Makaroun, M. S., Peitzman, A. B., Steed, D. L., and Webster, M. W. TcPO2 values in limb ischemia: Effects of blood flow and arterial oxygen tension. J. Surg. Res. 40: 482, 1986. 17. Lynch, T. G., Hobson, R. W., Wright, C. B., Garcia, G., Lind, R., Heintz, S., and Hart, L. Interpretation of Doppler segmental pressures in peripheral vascular occlusive disease. Arch. Surg. 119: 465, 1984. 18. Niinikoski, J. Oxygen and wound healing. Clin. Plast. Surg. 4: 361, 1977. 19. Greenhalgh, D. G., and Warden, G. D. Transcutaneous oxygen and Carbon Dioxide measurements for determination of skin graft ‘‘take.’’ J. Burn Care Rehabil. 13: 334, 1992. 20. Franzzeck, U. K., Galke, P., Bernstein, E. F., Jr., Golbranson, F. C., and Fronek, A. Transcutaneous PO2 measurements in health and peripheral arterial disease. Surgery 91: 156, 1982.

srga

AP: Surg Res