Green Light Photoplethysmography Monitoring of Free Flaps

Background  Monitoring strategies have been developed to address the issue of detecting postoperative free flap ischemia in an effort to permit intervention and flap salvage. No one existing noninvasive method has been widely accepted in a clinical setting. Green light photoplethysmography (GLP) uses a diode to transmit green light into a tissue. Reflected light from hemoglobin in dermal capillary red blood cells is analyzed as light intensity along a frequency spectrum. A pure peak signal (1-2 Hz) is identified and provides a way to distinguish between perfused and nonperfused tissue.

Design  Prospective, blinded comparison.

Subjects  Sixty of 72 consecutive patients considered for free flap reconstruction were enrolled in a protocol to evaluate the efficacy of GLP.

Intervention  After free flap elevation, but before pedicle ligation, 120-second baseline measurements were obtained; 120-second measurements then occurred 5 minutes after the onset or release of individual venous or arterial occlusion. Signals were processed by fast Fourier transfer; a mean alternating current–direct current (AC/DC) ratio was cultivated for each signal. All data were analyzed in a blinded fashion.

Results  The AC/DC ratio of GLP was statistically significant across all flap perfusion states (P<.001). Each condition resulted in a unique GLP signal within 5 minutes of manipulation of each vessel.

Conclusions  Green light photoplethysmography with AC/DC ratio analysis provides a rapid, precise method with which to determine flap ischemia and can differentiate venous compromised and arterial compromised flaps almost immediately after the onset of an ischemic insult. It may provide a clinically useful tool for postoperative free flap monitoring.

MICROVASCULAR free tissue transfer has emerged as the standard of care for reconstruction of complex head and neck and skull base defects after trauma or oncologic ablative surgery. Success rates have steadily improved to a level of 95% or greater in most series.¹ A critical factor in achieving a high rate of flap survival is the monitoring of the free tissue transfer after reimplantation to detect any changes that might indicate ischemia. The time interval for reestablishing vascular patency, however, is limited by the no-reflow phenomenon.² It does not exceed 8 to 12 hours based on animal studies and clinical observation.³ Prompt reoperation and correction of the microvascular obstruction are necessary to maintain flap viability. Revision is usually successful during this early critical period.^4,5

Monitoring strategies have been developed to address the issue of detecting postoperative flap ischemia in an effort to permit intervention and flap salvage.⁶ No one existing technique has been shown to provide an accurate and precise method for noninvasive monitoring of tissue perfusion in a clinical setting. The ideal technique should be noninvasive; reliable; objective; reproducible; economical; able to rapidly assess acute blood flow changes and to continuously monitor tissue perfusion; and easy for all medical personnel to use and interpret.⁷ Also, it should differentiate between arterial and venous compromise. None of the techniques currently available possesses all these qualities.

Green light photoplethysmography (GLP) uses a diode to transmit green light into a tissue. Reflected light from hemoglobin in dermal capillary red blood cells is analyzed as light intensity along a frequency spectrum. A pure peak signal (1-2 Hz) is identified and provides a way to distinguish between perfused and nonperfused tissue. Processing incoming signals by fast Fourier transfer (FFT) and the discrete cosine transform⁸ has allowed us to focus our examination at the physiologically significant frequencies of reflected light to the exclusion of physiologic and nonphysiologic sources of "noise." An initial pilot study that evaluated only radial forearm free flaps has demonstrated GLP to be accurate, reliable, and reproducible.⁹ We describe a series of free tissue transfers to demonstrate the efficacy of GLP free flap monitoring among diverse types of free flaps and improved signal analyses of the GLP.

All patients admitted to the University of Washington Otolaryngology–Head and Neck Surgery Service, St Louis, Mo, for head and neck reconstructive procedures were solicited for participation in this study between December 1996 and January 1998. The study and consent were reviewed and approved by the University of Washington Human Subjects Committee, and informed consent was obtained by the senior surgeon (N.D.F.) before surgery.

All flaps that were analyzed had a cutaneous portion available to which the GLP probe could be affixed with double-sided adhesive film (3M Co, Milwaukee, Wis). The prototype dual wavelength photoplethysmography transceiver (Cyber Medical Systems Ltd, Tampa, Fla) (Figure 1) was sterilely draped and brought into the operative field. Data were collected and archived on a laptop computer (Pentium Processor with 166 MHz, 32 megabytes of RAM, and >1 gigabyte hard drive). A commercially available software package (Vital Signs 5.1; Cyber Medical Systems Ltd) was used for data collection, storage, retrieval, real-time viewing, and signal processing (Figure 2). The alternating current–direct current (AC/DC) ratio is an index of reflected signal strength and was calculated for each GLP signal. This was done by exporting of data sessions from Vital Signs 5.1 as American Standard Code for Information Interchange (ASCII) files, which were analyzed using statistical software (SAS Institute Inc, Cary, NC).

A baseline measurement was performed in the area of the cutaneous portion of the planned free flap harvest after the area was prepared and draped but before any surgery and/or tourniquet inflation for extremity flaps was performed. Measurements were then taken after flap elevation, during flap reperfusion (after tourniquet release when applicable), but before vessel ligation and tissue transfer. At this point, the only connections of the flap to the donor site were the main flap artery and the draining vein.

Subsequent data were collected as follows: a raised flap baseline, 5 minutes of continuous venous occlusion, 5 minutes after the release of venous occlusion, 5 minutes of continuous arterial occlusion, and 5 minutes after the release of arterial occlusion. Occlusion was performed with a single microvascular clamp (V3 Acland; Accurate Surgical and Scientific Instruments Corp, Westbury, NY) on either vessel. No topical pharmacologic agents were applied to the pedicle vessels during or after occlusion. Measurements of 120 seconds were obtained for each of these conditions.

The participants were stratified to 3 groups based on the type of free flap: fasciocutaneous (radial forearm and scapula), osseocutaneous (fibula, iliac crest, and scapula), and myocutaneous (rectus abdominus and latissimus dorsi). The study parameters were analyzed within and across these 3 flap groups.

All data were stored in a secure hard drive, with patient data coded for anonymity for the duration of the project. All data were analyzed in a blinded fashion by a statistician to determine the statistical significance. Results from spectral analysis were compared with the known status of vascular pedicle. Repeated-measures analysis of variance was used for comparisons of matched series of data (between subgroups of free flaps). It was also used to determine if the means of the AC/DC ratios were different across flap states and flap sites. The Duncan multiple range test was used to determine if there were clusters of significance across flap states.

Sixty of 72 consecutive patients considered for free flap reconstruction were enrolled in a protocol to evaluate GLP monitoring. Twelve patients either declined enrollment in the study or did not ultimately require free tissue transfer. Sixty fasciocutaneous, osteocutaneous, and myocutaneous flaps were evaluated and are listed below:

Three hundred sixty individual samples were collected. Loss of adhesiveness of the GLP probe to the underlying skin created excessive motion, which resulted in excessive noise, and made 4 samples (0.9%) uninterpretable. These were excluded, which left 356 data samples suitable for analysis.

The original 356 samples contained 991,319 GLP data points. Observations with a signal-noise ratio of less than 3-dB threshold value were omitted. Omitting these observations was done as a filtering strategy to prevent noise from interfering with low-amplitude flap signals. Low-amplitude ratios were attributable, in part, to poor contact between the probe and the flap skin paddle. The net data set contained 686,856 observations across all flap sites. The AC/DC ratio was calculated for each signal.

The average AC/DC ratio was significantly different across all flap sites (P<.001). The mean AC/DC ratio is presented in Table 1. There was also a significant difference in means across flap states for each flap type: radial forearm (P<.001), rectus abdominus (P<.001), fibula (P<.001), iliac crest (P<.001), and latissimus dorsi (P<.001). Within each flap site, all flap states were significantly different (P<.001 for all sites) (Table 2).

Postoperative assessment or monitoring of vascular patency and tissue perfusion is an essential element in the successful outcome of microvascular free tissue transfer.^7,9 The success or failure of this technique relies principally on the patency of the microvascular anastomosis. The original technique for monitoring, which is still used currently, consists of observation of color, dermal bleeding, and capillary refill. This technique is subjective, labor intensive, and qualitative. Moreover, a great deal of judgment, which only comes through years of training and experience, is required in the interpretation of its results. At least 19 different techniques have been described so far,^6,7,10,11 which indicates that the ideal monitoring system is still to be determined. Laser Doppler flowmetry (LDF) is an alternative technology that has been in use over the past 2 decades as a means to monitor nutrient perfusion and is arguably the most accurate method in current use.^12-15

The advantages of LDF include its noninvasiveness, portable instrumentation, and the strong correlation to many invasive methods of tissue blood flow monitoring. Most authors have concluded that LDF has offered increased sensitivity in the clinical monitoring of perfusion, and some have claimed that it has resulted in improved survival of free tissue transfers.^12-14 The output of LDF is subject to a high degree of signal artifact. It can be affected by probe movement, changes in ambient conditions, fluid collection beneath the probe, and electrical noise.

In a previous report, Stack et al¹⁶ demonstrated that the LDF signal, when triggered to the cardiac cycle and averaged, will provide a method to accurately and precisely monitor nutrient perfusion of a microvascular free tissue flap. Moreover, this technique demonstrated characteristic and specific changes in signal with arterial and venous occlusion, respectively. An averaged signal with reduced artifact was produced by this method. The entire processes of averaging and waveform analysis can be automated and performed in minutes. Continuous flap monitoring and trend analyses, which are currently being used, are no longer necessary. This method permits timely intervention in cases of flap ischemia, with a resultant increase in salvage rate.

Although LDF has been shown in our studies to be precise and accurate in the detection of compromised perfusion, the hardware is quite expensive, approximately $15,000 per unit. The GLP device used in this study has the advantage of a superior light spectrum to LDF (575 vs 633 nm), which maximizes light reflectance from blood cells based on the photodiffusion theory. The instrumentation is far more economical than LDF (estimated preproduction cost, $500 per unit), which makes it a less expensive alternative for perfusion monitoring and may result in increased availability and use. Our results demonstrate that GLP monitoring can detect a perfusion abnormality in a very timely fashion, specifically, within 5 minutes of vascular compromise. This window often exceeds the time required to observe changes in perfusion by conventional methods. It also supercedes the likely window of time in which an ischemic flap is detected in a clinical setting of conventional intermittent flap inspection.

We evaluated each individual GLP signal by calculating the AC/DC ratio. Group means for AC/DC ratios were calculated for groupings based on flap type and perfusion state. The AC/DC ratio is an index of reflected signal strength. This index is proportional to the movement of corpuscles through capillary beds, which diminishes with manipulation of the tissue and ischemia (whether venous or arterial in origin). Arterial occlusion approaches 0, whereas venous occlusion is low but not 0 at 5 minutes; this is because there is still capillary corpuscle movement with ongoing arterial inflow and the venous circuit is acting as a capacitance vessel. There is a statistically significant difference between arterial and venous occlusion as measured by the AC/DC ratio, which, to our knowledge, has not been demonstrated by any other monitoring technique (P<.001). The postarterial occlusion signal may be relatively increased in some instances owing to the well-documented postischemic hyperemia phenomenon.

Given the current high rate of successful free tissue transfers, it will be difficult to make further significant reductions in flap ischemic complications by monitoring alone. The 5% to 25%^5,17 incidence of postoperative and microvascular anastomotic compromise dictates early recognition and timely reexploration and revascularization to salvage these flaps and to maintain a high overall success rate. In this study, we demonstrated that GLP has the potential to meet all the qualities of an ideal monitoring system. Improvements in probe design currently under way will maximize transmission and signal reception and will permit effective monitoring and study in the postoperative setting.

Accepted for publication November 17, 1999.

This study was supported in part by the American Academy of Facial Plastic and Reconstructive Surgery Investigator Development Award.

Presented in part at the Spring Meeting of the American Academy of Facial Plastic and Reconstructive Surgery, Palm Desert, Calif, April 28, 1999.

Corresponding author: Neal D. Futran, MD, DMD, Department of Otolaryngology–Head and Neck Surgery, University of Washington School of Medicine, 1959 NE Pacific St, Room BB1165, Box 356515, Seattle, WA 98195-6515.