A new method for the prediction of peripheral vascular resistance from the preoperative angiogram

Transcription

1 A new method for the prediction of peripheral vascular resistance from the preoperative angiogram Wayne W. LaMorte, M.D., James O. Menzoian, M.D., Anton Sidawy, M.D., and Timothy Heeren, Ph.D., Boston, Mass. The preoperative angiogram is widely used to estimate runoffprior to infraingninal bypass grafting, but the traditional method of angiographic scoring (, 1, 2, or 3 based on the number of patent tibial vessels) correlates poorly with measurements of peripheral vascular resistance. We assigned a score of, 1, or 2 to each of four parameters (anterior and posterior tibial arteries [AT and PT], peroneal artery [PER], and plantar arch [ARCH]) on the preoperative angiograms of 39 patients admitted for elective femoral bypass grafting. These scores were then examined for their ability to correlate with the peripheral vascular resistance measured in each patient at the time of surgery. Multiple linear regression suggested that the scores assigned to the AT, PT, and ARCH were significantly correlated with observed resistance, but the PER subscore was not. Multiple linear regression of the AT, PT, and ARCH subscores suggested that resistance could be predicted from the equation: In(Resistance [mm Hg/ml/min]) = (AT) -.25(PT) -.358(ARCH), for which r =.78 and p <.1. The ability of this relationship to predict resistance was tested by a division of the patient population into two roughly equal groups based on their observed resistance at surgery. When tested in this fashion, this equation had a sensitivity of 88% and a specificity of 82%. These findings suggest that the preoperative angiogram can be graded in a simple, yet unambiguous way, which allows a reasonable prediction of peripheral vascular resistance. (J VASe SUgG 1985; 2:73-8.) In addition to defining the location and extent of peripheral vascular disease, the preoperative angiogram is widely used as a means of estimating peripheral vascular resistance prior to bypass grafting. Traditionally, runoff has been graded as, 1, 2, or 3 based on the number of patent tibial vessels below the knee. Wc recently demonstrated, however, that this scoring technique correlates poorly with actual measurements of peripheral resistance in patients about to undergo femoral bypass grafting.~ The failure of the traditional scoring technique to predict runoff appeared to result from (1) imprecise criteria for deciding to grade a given tibial vessel as present or absent, (2) an inabiliw to distinguish poor vessels From the Department of Surgery (Drs. LaMorte, Menzoian, and Sidawy) and School of Public Health (Dr. Heeren), Boston University School of Medicine. Supported by a Biomedical General Research Support Grant from Universi~ Hospital (GRS NIH 2S7RR ). Reprint requests: Wayne W. LaMorte, M.D., Department of Surgery, Room L-98, Boston Universi~ School of Medicine, 8 East Concord St., Boston, MA from good ones, and (3) the assumption that each of the tibial vessels contributed equally to runoff: In the present study we established precise criteria for the assignment of a score of, I, or 2 to each of the following parameters seen on the preoperative angiogram: (1) the anterior tibial artery (AT), (2) the posterior tibial arte~ (PT), (3) the peroneal artery (PER), and (4) the plantar arch (ARCH). The subscores thus assigned to each of 39 consecutive patients were then analyzed for their abili~ to predict the peripheral vascular resistance measured at the time of surgery. METHODS Patient population. Patient studies were approved by the Internal Review Board at University Hospital, and each patient signed an informed consent. The patient population consisted of 39 consecutive patients admitted for elective infrainguinal bypass grafting. Each patient underwent preoperative angiography by standard procedures, which examined the arterial tree through the plantar arch. 73

2 74 LaMorte et al. ]ollrllil] I)]' VAS(It :LAF, St W,(;ERY (A) (B) (C) (D) AT 2 PT 2 P 2 A 2 PT P 2 A 1 (E) AT PT 2 ~! AT 2 ( P o /t/ A o A AT 1 vr P A werc studied as outlined ill a previous report. ~ Bricfly, systemic anticoagulation was produced by administration of heparin (1 mg/kg of body weight) given intravenously. Just prior to performance of the distal anastomosis, the distal arteriotomy was cannulated with a carotid artery shunt (Ar~,le, Sherwood Medical Company, St. Louis, Mo.) held in place with silicone rubber vessel loops. The artery was then per-. fused with sterile saline solution at constant (nonpulsatile) flow rates of 5, 75, and 1 ml/min tbr 15 seconds at each flow rate. Intra-arterial pressures were monitored continuously with a 128C pressure transducer (Hewlett-Packard Company, Waltham, Mass.) and were recorded on a strip chart recorder. During each infusion, pressure rapidly rose to a plateau and then remained constant until the infusion was stopped. Resistance was calculated at each flow rate by use of the equation: R = P/F Fig. 1. Examples of angiographic patterns and manner in which they were scored. A illustrates a normal angiogram that would receive 2 points for each of four subscores listed in Table I. B, Arch would receive only 1 point since it is incomplete and there is no filling of dorsalis pedis in foot. C, AT would receive 1 point since it extends only halfway down leg; PT and ARCH are absent and would each receive points. D would be assigned 2 points each for AT and PT and points for ARCH and PER, which were both absent. E would receive 2 points for ARCH, which fills through collaterals, but points for AT, IT, and PER. Scoring of preoperative angiograms. The angiographic scoring system used in this study was designed to meet the following criteria: (1) that it reflect not only the presence or absence of tibial vessels but also indicate whether the angiographic findings suggested that a given vessel was of poor or good quality; (2) that it indicate the degree of atherosclerotic disease in the most distal arterial runoff bed (ARCH); and (3) that the scoring system be unambiguous, yet simple enough to be easily applied. Accordingly, each of the three tibial vessels and the ARCH was scored as, 1, or 2 according to the criteria outlined in Table I. The application of these criteria to specific angiographic patterns is illustrated in Fig. 1. Measurement of peripheral vascular resistance. Patients undergoing infrainguinal bypass grafting in which P is the measured plateau pressure in mm Hg at a given flow rate, F is the flow rate in ml/min, and R is the resistance (mm Hg/ml/min). For each patient the calculated resistance was found to be comparable at each of the three flow rates. Therefore, to simpliff data analysis, the three resistances calculated for a given patient were averaged to obtain a mean resistance. RESULTS The angiographic subscores and the measured resistance of each of the 39 patients are shown in Table II. There was considerable variation in the extent of disease and in its pattern of distribution from patient to patient. Measured mean resistance ranged from a low of.22 to a high of 5. mm Hg/ml/min. Simple linear regression of the sum of the four subscores. Simple linear regression of the observed resistance as a function of the sum of the four subscores of each patient produced a correlation coefficient (r) of.61, which suggests a significant relationship that could be described by the following equation: R = [sum ofsubscores] (r =.61) (1) To test the predictive value of this relationship, the patients were divided into a high-resistance group and a low-resistance group based on the resistances observed at surgery. A resistance of 1. mm Hg/ml/ min was used as an arbitrary cutoff point simply because it conveniently divided the patient population into two roughly equal groups (Fig. 2). With

3 Volume 2 Number 5 September 1985 Prediction #'peripheral vascular resistance 75 Table I. Criteria used for scoring preoperative angiograms Score Criteria AT 2 Artcrv in continuity with poplitcal artery Traverses lower leg and enters lower leg at level of malleolus 1 Traverses lower leg only partially Not seen on angiogram 1Yr~ _ PER 2 In continuity with popliteal artery, Traverses lower leg to ankle mortise 1 Traverses lower leg only partially Not seen on angiogram ARCH 2 ARCH vessels in fbot completely filled PT and DP continuous with ARCH 1 ARCH vessels filled bv either DP or FF but not both No filling of ARCH AT = anterior tibial artery.; PT = posterior tibial arteq,; PER = pcroneal arteq,; ARCH = plantar arch; DP = dorsalis pedis artcq,. ~Scoring criteria ~.re same as for AT. Table II. Angiographic scores and observed resistance (mm Hg/ml/min) Subscores Observed Patient AT PER PT ARCH Subscore total resistance i I O

4 76 La3lorte et al. Journal of VASCUI.AR SURGERY 5. (D c 4. t~ w 3. > (/).a :F I I I Resistance Predicted from Equation 1 Fig. 2. Observed resistance as function of resistance predicted by simple linear regression of sum of four subscores. Horizontal and vertical lines divide patient population into four subsets and illustrate predictive value of equation 1. Of 17 patients observed to have resistance greater than 1., 14 were correctly predicted; 14 of 22 patients with observed resistance less than 1. were correctly predicted. If predictions had been perfect, they would coincide with diagonal line. Table III. Correlation coefficients (and probabilities) among angiographic subscores and mean resistance A T PER PT ARCH PER (.84) PT (.332)* (.4914) ARCH (.6464) (.682) (.9814) -- Resistance (.118)* (.433) (.77)* (.121)* in[resistance] *Statistically significant correlation. (.8)* (.52) (.5)* (.7)* this criterion to distinguish high-resistance patients from those with low resistance, equation 1 had a sensitivity of 82% and a specificity of 64%. Multiple linear regression of subseores. The relationship between the angiographic findings and resistance was further explored by an examination of the correlations between individual subscores and observed resistance. Table III shows the Pearson correlation coefficients and the statistical significance of each (in parentheses) for these parameters. Between subscores only the AT and PT subscores correlated significantly with one another. Subscores for the AT, PT, and ARCH each correlated significantly with observed resistance, but the PER subscore did not. The correlation coefficients were negative in each case, which suggests that a high score was associated with a low resistance. The significance of these correlations was improved by the natural logarithm of resistance (Table III, bottom). Consequently, subsequent analysis fo-

5 Volume 2 Number 5 September 1985 Prediction of peripheral vascular resistance q) c 4.,./) U~ 3 n.. 3. t...,.q I I I Resistance Predicted from Equation 2 Fig. 3. Observed resistance as function of resistance predicted by multiple linear regression of AT, IT, and ARCH subscores (equation 2). Of 17 patients with observed resistance greater than 1., 15 were correctly predicted; 18 of 22 patients with resistance less than 1. were correctly predicted. cused on the natural logarithm of observed resistance. A multiple linear regression was performed that related observed resistance (transformed to ln[r]) to the set of subscores. Again, the AT, PT, and ARCH subscores were significantly associated with resistance, but the PER subscore was not (Table III). Therefore, the PER subscore was dropped from the analysis, and observed resistance was regressed on the remaining three subscores. Based on this regression, the following relationship was suggested: In[R] = (2) [AT] -.25[PT] -.388[ARCH] (r =.78, p <.1) The abili~ of this equation to predict resistance from the angiographic subscores of the a9 patients is illustrated in Fig, 3. A resistance of 1. mm Hg/ ml/min was again used as a means of dividing patients with low resistance from those with high resistance. With this criterion equation 2 demonstrated a sensitivi~ of 88% and a specificity of 82%. DISCUSSION This stud), establishes the potential usefulness of the preoperative angiogram in the prediction of peripheral vascular resistance. In an earlier study we found that the usual method of scoring the angio- grams (, 1, 2, or 3 based on the number of patent tibial vessels) was a poor method of predicting the resistance measured at the time of surgery.~ The present study describes a new method of scoring and interpreting the preoperative angiogram that has greatly enhanced predictive value. The superiority of this new method probably resuits from two major factors. First, by the assignment of scores of, 1, or 2 to each of the tibial vessels and to the ARCH, it provides the oppomanity to describe the spectrum of vascular disease more completely. The traditional method forces the scorer to decide that a given vessel is either present (and presumably normal) or absent, but the technique described here allows the scorer to assign an intermediate value to each of the angiographic parameters. A second factor that probably contributes to the predictive value of this new method is that, by analysis of the data by multiple linear regression, the relative importance of each subscore can be assessed and used to weight the final estimate of predicted resistance. For example, in equation 2 the coefficients for the AT and the PT were nearly equal, which suggests that these two vessels contribute to runoff more or less equally. The larger coefficient for the ARCH suggests that its status has a greater impact on resistance than the AT or PT. In contrast, the PER had

6 78 Lazl4orte et al. VASCUI~AR SUI',Gt'iRY a coefficient close to zero and in fact did not correlate significantly with resistance. Consequently, the traditional method that evaluates the angiogram is hampered in a number of ways: (1) it wrongly assumes that each of the three tibial vessels has an equal impact on resistance; (2) it fails to take into account the quality of the ARCH; and (3) it offers no means of indicating that a vessel is present but partially compromised. The ability of this technique to predict peripheral vascular resistance is encouraging, but it should be stressed that equation 2 was derived in a retrospective fashion, and its ability to predict resistance needs to be tested prospectively. Moreover, although this technique appears to be capable of distinguishing patients with a resistance greater than 1. mm Hg/ ml/min from those with lower resistance, the relationship demonstrated in Fig. 3 is not perfect, particularly for patients with advanced atherosclerotic disease. Our ability to accurately predict resistance from the angiogram might be improved further by the use of a greater variety of grades for each anglographic parameter (e.g., to 5 rather than to 2), but it is also possible that important parameters have been totally overlooked. For example, equation 2 fails to take into account the presence of collateral branches. After much debate we intentionally omitted grading of collateral pathways for two reasons. The possibility exists that the presence (or abundance) of collateral circulation is itself a result of the severity of disease and thus becomes a confounding variable. Second, we were unable to agree on simple, objective criteria for scoring collateral pathways. Nevertheless, we acknowledge the possibili~ that collateral branches may have a significant impact on peripheral vascular resistance and/or graft patency. In contrast to this a priori exclusion of collaterals, thc PER was excluded from equation 2 because its score failed to correlate with observed resistance. The conclusion that the PER contributes nothing to runoffis difficult to accept, and we arc unable to account fbr its striking failure to correlate with resistance, particularly in view of the highly significant correlations obtained with the other tibial vessels. Despite these reservations these studies demonstratc the potential for accurately predicting peripheral vascular resistance from the preoperative anglogram. The abili~ to predict runoff preoperatively would be a useful contribution. Ascer et al. 2~ have demonstrated that outflow resistance is an accurate predictor of early graft failure, particularly in the case of femorodistal bypasses. The ability to predict resistance prior to surgery might alert the surgeon to a high-risk situation and stimulate a consideration of alternatives. Finally, if this scoring system can be validated, it would suggest a rational basis for the standardization of the grading of the preoperative angiogram. This would be particularly useful in the characterization of the degree of vascular diseasc among patients in various clinical studies. As such it would enable one to objectively weigh the merits of alternative therapeutic interventions. REFERENCES 1. Menzoian JO, LaMorte WW, Cantelmo NL, Doyle J, Sidawy AN, Savenor A. The pre-operative angiogram as a predictor of peripheral vascular run-off. Am J Surg (In press.) 2. Ascer E, Veith FJ, Morin L, White-Flores SA, Scher LA, Samson RH, Weiser RK, Rivers S, Gupta SK. Quantitative assessment of outflow resistance in lower extremity, arterial reconstructions. J Surg Res 1984; 37: Ascer E, Veith FJ, Morin L, Lesser ML, Gupta SK, Samson RH, Scher LA, White-Flores SA. Components of outflow resistance and their correlation with graft patency in lower extremiqr arterial reconstructions. J VAsc SUItG 1984; 1: