Redistribution of Myocardial Blood Flow With Topical Nitroglycerin in Patients With Coronary Artery Disease

Methods

Image Analysis

With the aid of a computerized coordinate system developed in our laboratory, each set of transaxial images was realigned along the longest base-to-apex axis to display images in both sagittal and anteroposterior planes. From these sets, cross-sectional images perpendicular to the long axis were reconstructed. Using the junction of the right ventricular wall with the anterior interventricular septum as a landmark, the images were sectioned circumferentially into 36 equal segments for each of 6 base-to-apex equidistant slices for a total of 218 sectional pies (Fig 1⇓). The mean count density, expressed as counts · pixel⁻¹ · second⁻¹, was computed for each myocardial sector.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Schematic of myocardial sectors. The cross-sectional left ventricular images were circumferentially sectioned into 36 equal pies for each of 6 base-to-apex equidistant slices. The mean values for [¹³N]ammonia and fluorodeoxyglucose radioactivity were then computed for each sector.

Global and Regional Myocardial Perfusion Measurements

Regional perfusion expressed as ¹³N retention was calculated for each of the 216 sectors (36 pies×6 slices). Global myocardial perfusion was determined as the average of all ¹³N retention measurements. Regional myocardial perfusion was defined as the average ¹³N retention for selected segments containing at least five contiguous sectors incorporating two adjacent slices. The regions designated as reduced perfusion, or “ischemic,” were defined as a reduction in ¹³N counts by at least 2 SDs below the mean of the maximum count density in the normally perfused or “nonischemic” sectors. The latter was defined as the five contiguous sectors incorporating two adjacent slices that contained the maximum count density. These nonischemic zones always corresponded to myocardial regions subtended by patent coronary arteries as seen angiographically. The ischemic zones were also required to demonstrate no change or an increase in FDG uptake (mismatch).

In the glucose-loaded state, ¹⁸FDG myocardial radioactivity was derived from counts · pixel⁻¹ · second⁻¹ of the last frame normalized for the blood radioactivity at 20 minutes. Accepting the fact that glucose utilization rates are highly variable, even in normally perfused myocardium, a region characterized as an infarct zone or scar had to show a net reduction both in FDG radioactivity (>2 SDs below the mean counts in zones of maximal FDG uptake) and a corresponding reduction in ¹³N retention as defined above for ischemic zones. This is illustrated in Fig 2⇓, in which normally perfused zones are seen alongside zones of reduced perfusion and scar in the same patient.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Linear display of circumferential slice No. 5 in a patient with a previous apical infarct (sectors 12-20) and a zone of posterolateral ischemia (sectors 25-35). Note the mismatch between the baseline ¹³NH₃ retention (○—○) and the ¹⁸F-deoxyglucose uptake (---) for sectors 25-35; nitroglycerin (—+—) preferentially enhanced flow to this ischemic zone, whereas both ¹³NH₃ and ¹⁸F-deoxyglucose were reduced in the zone of infarction.

Results

The placebo and nitroglycerin groups were comparable in terms of demographic, clinical, and angiographic characteristics (Table 1⇓). Patients in both groups experienced moderate effort-induced angina, with Canadian Cardiovascular Society grades of 2.1±0.3 and 2.3±0.4 for the placebo and nitroglycerin groups, respectively. The extent of coronary artery stenoses and left ventricular dysfunction was also similar in both groups. Most patients had more than one coronary vessel containing stenotic lesions of at least 50%. There were no differences in the amount or type of cardiac medications prescribed for each group. All patients were taking either long-acting oral or topical nitroglycerin prior to the study. All cardiac medications including nitroglycerin were discontinued at least 48 hours prior to the study. One patient took a sublingual nitroglycerin tablet (0.3 mg) on the morning of his study, but the 2,3-dinitroglycerin plasma level in this patient was undetectable by the time the [¹³N]ammonia acquisition scan was performed.

To illustrate the flow and metabolic characteristics of different zones in multivessel disease, a linear display of a circumferential sector from one of the patients is shown in Fig 2⇑. This patient had an anteroapical infarct (sectors 12 through 20), wherein is seen a reduction in both ¹⁸FDG and ¹³NH₃ retention values and between baseline and post–nitroglycerin patch values. Note is made of the mismatch between the elevated ¹⁸FDG and depressed baseline ¹³NH₃ values in the posterolateral zone of reduced perfusion (sectors 25 through 35). Nitroglycerin appears to have redistributed flow to the ischemic zone.

Reproducibility and stability of the perfusion method was analyzed by comparing the sector count densities for ¹³N retention between baseline and placebo states (Table 2⇓). The mean ¹³N retention values were 0.61±0.14 and 0.62±0.12 min⁻¹ for baseline and placebo, respectively (Tables 2⇓ and 3⇓). Although the intergroup values were almost identical, the interpatient perfusion values varied widely, with coefficients of variation of 22.9% and 19.4% for baseline and placebo groups, respectively, and an intraclass correlation coefficient of .51. There was, however, remarkable reproducibility within patients from one state to the next, with an average percent difference of mean count density of only 2.70±1.42. The average global ¹³N retention for both states was 0.62 min⁻¹ with confidence limits from 0.47 min⁻¹ to 0.77 min⁻¹ at the 95% level.

Fig 3A⇓ shows an example of the reproducibility of ¹³N retention data expressed as a percent change from baseline 3 hours following application of a placebo skin patch. The change from baseline was less than 5% for all sectors and slices. By contrast, Fig 3B⇓ shows considerably more variability in a patient who had received a nitroglycerin patch; the significant increase in ¹³N retention in the anterolateral zone (sectors 28 through 35) corresponds to a high-grade circumflex artery stenosis. The apparent decrease in perfusion of sectors 5 through 15, a zone identified as subtending a patent left anterior descending artery, may represent either redistributed flow attributed to the effect of nitroglycerin on conductance vessels or a decrease in wall stress and left ventricular end-diastolic pressure due to the unloading effect of nitroglycerin.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Line graphs. A, Example of the reproducibility of regional ¹³N retention images for a placebo study. Each line represents the percent change in ¹³N retention from baseline to placebo patch for each of six myocardial slices. Note that myocardial perfusion in any sector changes less than 5% for every cross-sectional slice of the left ventricular myocardium. B, Example of regional blood flow changes with topical nitroglycerin. Lines are as in Fig 3A⇓. Note the increase in ¹³N retention in sectors 27-35. This zone corresponded to an area of anterolateral ischemia documented by coronary angiography. Sectors 5-15 correspond to an anteroseptal zone supplied by a patent nonstenotic left anterior descending artery.

The relation between global myocardial perfusion as measured by the average ¹³N retention counts in all sectors and the work of the heart expressed as the double rate-pressure product is shown in Table 3⇑. Under resting supine conditions, there were no significant changes in either heart rate or systolic blood pressure for either group. If anything, there was a slight but insignificant decrease in the average peak systolic blood pressure with nitroglycerin, from 127±10 to 119±11 mm Hg, 3 hours following patch application. Although the baseline values for global ¹³N retention appear to differ between placebo and nitroglycerin groups (0.61±0.14 versus 0.75±0.22 min⁻¹), the variability between patients was high, with a coefficient of variation ranging from 13% to 29%. This interpatient variation in myocardial perfusion contrasts sharply with the intrapatient reproducibility of the ¹³N retention taken 3 hours apart for the placebo group (Table 2⇑ and Fig 3A⇑). Neither the global perfusion as expressed by ¹³N retention nor the double rate-pressure product was altered significantly with nitroglycerin. External work per unit flow, roughly expressed as the ratio of the double rate-pressure product to ¹³N retention, was unchanged by the application of either placebo or nitroglycerin patch.

The time-dependent changes in the double rate-pressure product with either nitroglycerin or placebo are illustrated in Fig 4⇓. Superimposed is the time course of plasma levels for 1,2-dinitroglycerin. Despite a progressive increase in plasma levels of nitroglycerin, the double rate-pressure product remained constant and, as a consequence, so did the requirement for myocardial perfusion (Table 3⇑). Plasma levels for 1,3-dinitroglycerin were also assayed and showed a similar time course.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Time curves of the double rate-pressure product (RPP) for both groups and 1,2-dinitroglycerin blood levels for the nitroglycerin (GTN) group (grp). Vertical axis is either nanograms/milliliter for 1,2-dinitroglycerin plasma levels or (mm Hg×bpm) · 100⁻¹ for the RPP. Vertical bars show SD for nitroglycerin group only. The coefficient of variation for the placebo group was less than 5%. There were no detectable plasma levels of 1,2-dinitroglycerin for the placebo group.

In sharp contrast to the lack of change in global myocardial perfusion, the topical application of nitroglycerin significantly altered the regional distribution of myocardial perfusion. Comparison of regional ¹³N retention between normal zones of perfusion and zones of reduced perfusion (ischemic segments) is shown in Table 4⇓. There were no changes in regional perfusion to either nonischemic or ischemic zones following topical application of the placebo patch. However, nitroglycerin appeared to preferentially increase perfusion to ischemic zones by increasing the relative myocardial retention of [¹³N]ammonia, from 73±13% to 94.9±18% (P<.05). As expected, proportional uptake of ¹³N radioactivity in the ischemic zone was, on average, less than in the nonischemic zone for both placebo and nitroglycerin groups at baseline (overall mean difference, −0.17±0.16 min⁻¹; P<.002). The preferential enhancement of flow to the ischemic zones with nitroglycerin is illustrated in Fig 5⇓. There was less than a 5% change in regional perfusion following application of the placebo patch regardless of which zone was analyzed. Similarly, the apparent decrease in regional perfusion to the nonischemic zone following nitroglycerin was less than 5% (P=NS). A change in radioactive uptake of less than 5% with any intervention is within the experimental error of the technique. On the other hand, there was a significant 11.3±3% increase in ¹³N retention in the ischemic zone following topical application of nitroglycerin.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Bar graph of percent change in regional myocardial perfusion (expressed as net retention of [¹³N]ammonia) from baseline to 3 hours postpatch for both “nonischemic” (normally perfused) and “ischemic” (reduced ¹³N retention) zones. Note that regional myocardial perfusion changed less than 5% in all instances except for the effect of nitroglycerin on blood flow to the ischemic zone. Hatched bars indicate nitroglycerin values; solid bars, placebo values; and vertical lines, SD.

Fig 6⇓ illustrates two polar maps of ¹³N radioactivity in a patient who received topical nitroglycerin. The left hand map (prepatch) shows the variation in regional uptake of [¹³N]ammonia throughout the left ventricle with reduced perfusion in the anteroapical and posterior segments (areas in blue). Three hours after application of nitroglycerin, the anteroapical segments (just above the center zone) appear to have filled in, with little or no change elsewhere. This example is a visual illustration of preferential perfusion to an ischemic zone with nitroglycerin under resting supine conditions.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Polar maps showing regional myocardial distribution of [¹³N]ammonia in a patient who received a nitroglycerin patch. Anatomic regions correspond to the cross-sectional graphic in Fig 1⇑. Baseline image at left shows a large zone of reduced perfusion (blue). Note that the anteroapical area fills in after nitroglycerin administration.

Discussion

This study demonstrated that topical nitroglycerin (0.4 mg/h) increased the myocardial retention of [¹³N]ammonia in ischemic zones by more than 11% while little or no change in nonischemic or global myocardial perfusion occurred. Placebo skin patches had no effect on either regional or global perfusion as determined by ¹³N retention. These dynamic adjustments in regional blood flow with nitroglycerin occurred in the supine rest state, where little or no change in left ventricular work, as determined by the double rate-pressure product, was observed. The ability to study the heterogeneous effects of nitroglycerin on the coronary vasculature of humans had to await quantifiable measurements of global and regional blood flow. PET offers a unique approach to examine and quantify regional coronary blood flow and, thereby, possible mechanisms of drug action, but it is not without limitations and requires careful scrutiny.

Myocardial zones of reduced perfusion were designated as “ischemic” and arbitrarily defined as a contiguous set of five sectors in two adjacent slices having average count · pixel⁻¹ · second⁻¹ of retained [¹³N]ammonia at least 2 SDs below the average counts in nonischemic zones. These measurements were done in the rest supine state. The data would perhaps have been more persuasive had the patients been subjected to exercise in order to induce ischemia. However, the use of exercise with PET is fraught with problems, not the least of which is exact repositioning of the heart in the tomograph scanner.¹³ Furthermore, interpretation of myocardial blood flow as an expression of [¹³N]ammonia extraction (or retention) by the myocardium is hazardous since any increase in net extraction in a high-flow state may seriously underestimate the true changes in myocardial perfusion.¹⁶ These drawbacks were mitigated by the combined use of ¹³NH₃ and ¹⁸FDG, which enabled us to identify and dissociate zones of ischemia from infarcted tissue.

Validation of our methods was supported by two observations. First, there was less than a 5% variation in ¹³N retention in any of the 216 sectors when comparing baseline with 3 hours of placebo patch (Table 2⇑ and Fig 3⇑). Second, there was over 90% concordance between a significant decrease in ¹³N retention of a myocardial sector (>2 SDs below the mean) and the presence of a significant stenosis (>50%) in the artery supplying that zone.

Instead of relying solely on static extraction data, regional perfusion was expressed as ¹³N retention derived from dynamic acquisition data. This provides more reliable and accurate estimates of flow.^{13 17} Contamination of myocardial retention data by partial volume and spillover effects becomes especially relevant when estimating radioactivity in areas of myocardial infarct thinning. This was partially avoided by calculating the arterial input function from the center of the ventricular lumen of the most basal segment, which minimizes spillover but does not eliminate it entirely. We were encouraged, however, by the consistent reproducibility of the ¹³N retention images before and after placebo. In all 10 patients receiving the placebo patch, the percent difference of the means for global and regional counts varied by less than 5% between baseline and placebo states.

The average ¹³N retention value for global ¹³N retention (Table 3⇑) appears less than some of the average regional perfusion values (Table 4⇑). This is explained by the wide regional variation in [¹³N]ammonia uptake in hearts with multivessel disease containing several regions of reduced perfusion and previous myocardial infarcts. Moreover, the baseline global retentions for the placebo group (0.61±0.14 min⁻¹) were, on average, actually less than the nitroglycerin group (0.75±0.22 min⁻¹). However, the interpatient variation was quite high (coefficients of variation, 20% and 22%, respectively, for the placebo and nitroglycerin groups), which is not surprising considering the variability of segmental disease and myocardial scar in these patients. On the other hand, the intrapatient reproducibility for ¹³N retention was very narrow for the placebo group, and it was the percent change in relative myocardial perfusion that was measured, using each patient as his or her own control. Furthermore, the baseline global value for the nitroglycerin group (0.75±0.22 min⁻¹) was within the confidence intervals of the values derived from both the baseline and placebo states (Table 2⇑).

The double rate-pressure product was used as an expression of left ventricular work in this study. This measurement ignores the contribution of left ventricular volume and wall stress as major determinants of oxygen demand and, hence, flow. However, it is well known that nitroglycerin reduces left ventricular filling volume,^{18 19} thereby maintaining or even reducing demand for global coronary blood flow in the face of an unchanged double rate-pressure product. Moreover, a decrease in left ventricular end-diastolic pressure by nitroglycerin would, theoretically, potentiate transmural blood flow, but this effect in the absence of a change in energy demand would be negligible. Moreover, it would be uniform for all segments.

As for the effect of nitroglycerin on the coronary circulation, it has been repeatedly observed that nitroglycerin visibly dilates the large epicardial branches of coronary arteries.²⁰ Moreover, there is oftentimes evidence for relief of large vessel coronary spasm within minutes of sublingual nitroglycerin²¹ as well as vasodilation of stenotic segments.²² What is less well known is the site and mechanism of the complex effects of this drug on coronary dynamics. While nitroglycerin dilates large epicardial coronary arteries, it has little or no effect on total coronary blood flow or coronary vascular resistance.^{6 7}

In a classic series of experiments Fam and McGregor^{8 23} compared the site and action of different coronary artery vasodilators. Using open-chest anesthetized dogs, they discovered that flow in the poststenotic coronary artery was enhanced with intravenous nitroglycerin by virtue of the relaxing effect of the drug on large conductance vessels. With no change in arteriolar resistance of the nonobstructed vascular bed, they postulated that coronary flow was preferentially directed through collaterals to the poststenotic zone of ischemia. By contrast, intravenous dipyridamole caused a significant reduction in coronary arteriolar resistance, thereby augmenting coronary flow to nonischemic myocardium at the expense of flow to the already dilated poststenotic vessel. Others have since substantiated the differential site of action of nitroglycerin on the coronary vasculature in experimental animals^{24 25 26} and humans.¹¹

Kurz et al²⁷ recently proposed an interesting and persuasive explanation for this heterogeneous coronary vascular response to nitroglycerin. The coronary arterioles apparently lack the necessary sulfhydryl-containing compound (l-cysteine) to convert nitroglycerin to its active metabolite, S-nitroso-l-cysteine.^{28 29} By suffusing epicardial microvessels in vivo with nitroglycerin, Kurz et al²⁷ were able to show significant dilatation of the arteriolar vessels only in the presence of a thiol, l-cysteine. This was a stereospecific reaction, suggesting an intracellular or membrane-associated mechanism. Hence, the lack of available sulfhydryl groups in the small resistance coronary vessels may explain the failure of nitroglycerin to alter small vessel coronary vascular resistance while dilating the large epicardial vessels.

However, it is possible that the observed increase in ¹³N retention to ischemic zones may in part be due to improved regional wall motion or a favorable transmural pressure gradient due to a decrease in left ventricular end-diastolic pressure and wall stress. An alternative explanation is that the so-called redistribution was actually the consequence of a direct effect of nitroglycerin on increasing the luminal diameter of vessel stenosis.^{10 22} This effect was at a dose that may or may not be achievable by topical nitroglycerin. Such a direct increase in cross-sectional stenosis area would, conceivably, alter global myocardial blood flow, assuming no change in flow to nonischemic zones. Although such an effect was not seen in this study, a mechanism of a direct vasodilatory effect cannot be refuted by the present results. Fujita et al,¹¹ also using computerized quantitatative coronary arteriography, found that intracoronary nitroglycerin dilated the recipient coronary arteries more than the donor arteries, suggesting an enhancement of flow through collateral channels.

In summary, our study shows that the application of topical nitroglycerin appears to exert changes in myocardial perfusion by preferential distribution of flow to areas of myocardial ischemia with little or no significant change in either total myocardial perfusion or cardiac work. This observation may help explain part of the antianginal properties of the coronary vasodilatory effects of nitroglycerin. The study also draws attention to PET as a potentially useful tool in unraveling the mechanisms of various interventions on regional myocardial blood flow.