Quantification of the spatial distribution of rectally applied surrogates for microbicide and semen in colon with SPECT and magnetic resonance imaging

Abstract

AIMS

We sought to describe quantitatively the distribution of rectally administered gels and seminal fluid surrogates using novel concentration–distance parameters that could be repeated over time. These methods are needed to develop rationally rectal microbicides to target and prevent HIV infection.

METHODS

Eight subjects were dosed rectally with radiolabelled and gadolinium-labelled gels to simulate microbicide gel and seminal fluid. Rectal doses were given with and without simulated receptive anal intercourse. Twenty-four hour distribution was assessed with indirect single photon emission computed tomography (SPECT)/computed tomography (CT) and magnetic resonance imaging (MRI), and direct assessment via sigmoidoscopic brushes. Concentration–distance curves were generated using an algorithm for fitting SPECT data in three dimensions. Three novel concentration–distance parameters were defined to describe quantitatively the distribution of radiolabels: maximal distance (D_max), distance at maximal concentration (D_Cmax) and mean residence distance (D_ave).

RESULTS

The SPECT/CT distribution of microbicide and semen surrogates was similar. Between 1 h and 24 h post dose, the surrogates migrated retrograde in all three parameters (relative to coccygeal level; geometric mean [95% confidence interval]): maximal distance (D_max), 10 cm (8.6–12) to 18 cm (13–26), distance at maximal concentration (D_Cmax), 3.8 cm (2.7–5.3) to 4.2 cm (2.8–6.3) and mean residence distance (D_ave), 4.3 cm (3.5–5.1) to 7.6 cm (5.3–11). Sigmoidoscopy and MRI correlated only roughly with SPECT/CT.

CONCLUSIONS

Rectal microbicide surrogates migrated retrograde during the 24 h following dosing. Spatial kinetic parameters estimated using three dimensional curve fitting of distribution data should prove useful for evaluating rectal formulations of drugs for HIV prevention and other indications.

WHAT IS ALREADY KNOWN ABOUT THIS SUBJECT

• Rectally applied drugs have been imaged using radioisotopes and magnetic resonance contrast agents. However, prior studies have not described the distribution and clearance of rectally-applied drugs in quantitative terms with respect to complex three dimensional paths through the gastrointestinal tract. Such tools would allow statistical comparisons of candidate products in development or comparison of drug product with the distribution of a drug target, for example, HIV infected seminal fluid.

WHAT THIS STUDY ADDS

• New quantitative spatial parameters, derived from three dimensional curve fitting, have been successfully applied in this study to quantify the distribution of rectally-applied gels. Indirect assessment using nuclear medicine techniques avoided the distortion and redistribution associated with sigmoidoscopic sampling. Thus, these measurements can be repeated over time to create concentration–distance–time surfaces to describe rectal product distribution and clearance within the gastrointestinal lumen to inform microbicide and other rectal product development.

Introduction

Microbicides are vaginally and rectally applied drugs intended to prevent sexual transmission of HIV infection. Two studies of tenofovir gel applied vaginally yielded modest or no reduction of HIV infection [1, 2]. Several large efficacy studies of vaginal formulations are currently underway. However, there have been comparatively fewer studies of microbicides for rectal use. Receptive anal intercourse is the primary route of HIV infection in men who have sex with men (MSM) and a very important route for HIV transmission in women [3–7], yet few studies have been done to evaluate microbicides in MSM [8].

An optimal rectal product would achieve adequate concentrations to prevent HIV infection in all areas of the distal colon exposed to HIV-infected ejaculate of an insertive sexual partner and subsequently remain within that distribution as long as the virus is present. Design of such a product requires quantitative methods to describe the distribution and clearance of both drug and virus in the distal colon. Depending on the mechanism of action of the candidate microbicide, either luminal or tissue concentrations may be of greater relevance. However, because luminal distribution likely influences subjacent tissue concentrations, luminal concentrations are relevant even for drugs active in tissue.

Quantitative assessment of the colonic distribution and clearance of rectal microbicide candidates will guide optimization and inform selection of candidates to go forward in development. Application of these quantitative methods to clinical studies should improve the rational design and development of rectal microbicide products. Demonstration of efficacy of these rectal products will have substantial public health benefits in preventing HIV infection.

Quantitative methods to assess distribution and clearance of drugs in the lumen of the colon, however, are not well developed. We recently demonstrated the feasibility of describing drug distribution in the distal colon using non-invasive imaging with single photon emission computed tomography (SPECT)/computed tomography (CT) and magnetic resonance imaging (MRI), and invasive colon biopsies [9]. We wanted to develop more refined quantitative methods to summarize colonic drug distribution in an effort to provide pharmacokinetic parameter estimates to describe drug behaviour in this anatomical compartment. We carried out this study to develop spatial kinetic parameters to describe the spread of radiolabelled surrogates for rectally-applied microbicide gels and semen.

Methods

Study overview

This open label study had three parts and two phases for a total of six different test conditions (Table 1). The study was designed to examine the distribution of a microbicide gel followed by simulated intercourse (part I), semen surrogate following simulated intercourse (part II), or microbicide surrogate followed by simulated intercourse and then semen surrogate dosing (part III). Different methods of assessing surrogate distribution were used. In phase A, data were acquired using SPECT/CT and MRI. In phase B, SPECT/CT and sigmoidoscopy were used. Subjects were followed for 24 h after each rectal dose was administered. During phase A, SPECT/CT was performed immediately (within 30 min) after dosing and then at 4, 10 and 24 h after dosing. MRI imaging was obtained following each SPECT/CT scan at 1, 5, 11 and 25 h following dosing. In phase B, each subject underwent SPECT/CT imaging 4 and 24 h after dosing. Flexible sigmoidoscopy with luminal brushing and biopsy followed each SPECT/CT scan at 5 and 25 h after dosing.

Table 1.

Protocol schema. Six test conditions are explored across three parts (defined by presence of microbicide surrogate, coital simulation, and semen surrogate) and two phases (defined by methods of observation, each including SPECT/CT, but one also includes MRI and the other also includes direct sigmoidoscopic sampling). The sequence listed within each study part is repeated for each of the two phases. ‘Microbicide’ and ‘semen’ refer to surrogates for each with radiolabelling as described in the text

	Imaging – Phase A (Post dose SPECT/CT and MRI)	Luminal sampling – Phase B (Post dose SPECT/CT and sigmoidoscopy)
Part I	Microbicide → Coital simulation → No semen	Microbicide → Coital simulation → No semen
Part II	No microbicide → Coital simulation → Semen	No microbicide → Coital simulation → Semen
Part III	Microbicide → Coital simulation → Semen	Microbicide → Coital simulation → Semen

Study participants

Following Institutional Review Board approval, eligible subjects were informed of the study purpose, procedures and risks. After providing written informed consent, subjects were enrolled. Subjects were eligible if they identified as men who have sex with men (MSM) and had a history of receptive anal intercourse and rectal use of cellulose-based sexual lubricants. Subjects were excluded from enrolment if they had any laboratory evidence of a coagulopathy, history of anorectal surgery or painful anorectal conditions or active gastrointestinal disease that might affect distribution of the gel vehicle or otherwise limit their study participation. Subjects were also excluded if they had a known contraindication to any study procedure.

Study products and dosing

Microbicide vehicle surrogate

The microbicide vehicle surrogate was 10 ml of K-Y Jelly™ (Advanced Care Products, Raritan, NJ, USA). For imaging in the microbicide surrogate alone study, 1000 µCi of ^99mTc sulphur colloid and a 1:100 dilution of gadolinium-DTPA were mixed into 10 ml of lubricant gel for SPECT and MRI, respectively (part I). For the combined microbicide surrogate and semen surrogate phase (part III) the K-Y Jelly was combined with 250 µCi of ¹¹¹Indium-DTPA rather than ^99mTc-sulphur colloid. This allowed the use of ^99mTc-sulphur colloid for radiolabelling the concomitant semen surrogate. Each isotope has discrete, largely non-overlapping emission energies allowing simultaneous imaging.

Semen surrogate

The semen surrogate vehicle was a combination of two commercially available personal lubricants (K-Y Jelly™ and FemGlide®, Cooper Surgical, Trumbull, CT, USA) combined in a ratio to match physiologically the osmolality and viscosity of semen. The semen surrogate vehicle (both part II and III) consisted of 5 ml total volume with 1000 µCi of ^99mTc sulphur colloid incorporated. ^99mTc sulphur colloid has an average particle diameter of 100 nm, similar in size to HIV particles.

Dosing and coital simulation

The microbicide surrogate (part I and II) was dosed via a rectal luer adaptor (product 35–1107; Professional Compounding Centers of America, Houston, TX). This was followed by coital simulation using a phallic-shaped vaginal dilator (cylindrical shaft with conical end) composed of a flexible, latex-free silicone that is an FDA-approved device for human use (Vaginal-Hymenal Silicone Dilator, Milex Products, Chicago, IL, USA). While in the recumbent position, subjects inserted the lubricated vaginal dilator into the rectum and manipulated the dilator to simulate receptive anal intercourse (5 min of 1 cycle per second in/out cycles). This coital simulation algorithm was based on MSM focus groups previously described by our group [10]. In studies where the semen surrogate was administered (part II and part III) the coital simulation was performed using a vaginal dilator modified to allow dosing of the semen surrogate through the dilator by means of an intravenous catheter (Arrow International, Reading, PA) inserted into the anatomic position of the urethra. The semen surrogate (parts II and III) was administered through the catheter during coital simulation, followed by 10 additional in/out cycles of the dilator.

Study procedures

Subjects were admitted to the General Clinical Research Center (GCRC) at The Johns Hopkins Hospital for each study phase. On the evening of admission, subjects received a clear liquid diet and then fasted except for water after midnight. Subjects received a bowel preparation prior to dosing on day 1 consisting of either a 250 ml tap water enema (phase A) or Fleet Phospha-Soda oral solution (phase B). In the morning of day 1, a microbicide surrogate was administered per rectum (part I and III), followed by coital simulation (parts I, II and III), and semen surrogate dosing (parts II and III). The dilator was then removed and measured for radioactivity using a dose calibrator (CRC 15-W, Capintec, London, Ontario, Canada). Study subjects remained supine until the 4 h imaging was completed. If the subject had a bowel movement during admission, the stool was collected and measured for gamma activity. Subjects were discharged after the 24–25 h imaging assessments were completed. Subjects were questioned about adverse events and underwent a symptom-directed physical examination prior to discharge. Phase B was initiated at least 7 days after phase A was completed. After admission and bowel preparation on day 0 and the study-specific dosing sequence on the morning of day 1, SPECT/CT imaging was performed at 4 h post dosing. Following SPECT/CT, subjects were transferred by stretcher to the endoscopy suite where they underwent endoscopic sampling via flexible sigmoidoscope to 60 cm (model CFQ160S, Olympus America, Center Valley, PA). At 10 cm intervals, two separate mucosal brushings were obtained using a cytology brush (model 60315, Ballard Medical Products, Draper, UT, USA). The brush was introduced through the endoscope port, extended from the catheter sheath and swept against one side of the colonic wall for luminal sampling. The brush was withdrawn into its sheath and removed from the sigmoidoscope. The brush was re-extended from the sheath and clipped off at the neck into a pre-weighed 4 ml sterile vial for gamma counting with correction for weight and radioactive decay. This procedure was repeated to sample the opposite side of the colonic wall. Subjects remained supine up to 6 h after microbicide administration (until completion of the first endoscopic procedure). SPECT/CT and endoscopic sampling were repeated 24 h post dosing. All stools were collected, weighed and a 1 ml aliquot measured for gamma activity. Once all endoscopic sampling had been obtained, a 500 ml tap water enema was administered to wash out residual radionuclide. Subjects were questioned regarding any adverse events and then discharged from the inpatient unit.

SPECT/CT imaging

SPECT/CT imaging was performed as previously described [9] with a dual-head VG series system (GE Medical Systems, Waukesha, WI, USA) equipped with a low-end CT unit (Hawkeye). Summarizing, SPECT and CT images were reconstructed using filtered back projection with ordered subset expectation maximization (OSEM) into a 128 × 128 matrix size and fused in the General Electric eNTEGRA workstation (version 1.04, GE Medical Systems, Waukesha, WI, USA). The image data were exported in dicom file format, attenuation corrected, later converted to the Mayo Analyze image file format (*.img *.hdr pair) using MRIcro, a freeware program developed by Chris Rorden at University of South Carolina http://cnl.web.arizona.edu/mricro.htm). The Analyze files were read into R programming environment (R version 2.5.1. [11]) with the library of AnalyzeFMRI.

Dual isotope imaging

To correct for crosstalk between ¹¹¹In and ^99mTc images, downscatter from ¹¹¹In to the ^99mTc energy window was estimated and compensated using a model-based method [12]. Briefly, the scatter in the object is modelled using the effective source scatter estimate technique, including contributions from both ¹¹¹In and ^99mTc photo peaks [13]. Photon interactions inside the collimator-detector system, including the penetration and scatter components, are estimated using pre-computed tables calculated from Monte Carlo simulations. The estimated downscatter is then compensated for during iterative OSEM reconstruction of the ^99mTc images by adding the downscatter estimate to the computed projections at each iteration [14].

Quantitative three dimensional curve fitting

Principal curve algorithm to quantify the spread of surrogate

After the image files containing SPECT data were read into R, the gamma signal strength of ^99mTc or ¹¹¹In were viewed in a 3-D array representing the anatomical location: sagittal (X from right to left), coronal (Y from back to front), and transverse (Z from head to foot). To construct a 3-D curve through the gamma emission signal, a novel principal curve algorithm was developed and described in detail [15]. Briefly, the three dimensional position of a curve, f, through the colon at t ∈[0, 1] was defined by:

where ψ represents X, Y and Z, were knots placed at equally spaced quantiles of {t_i}, K + 4 was the degree of freedom of the smoother and σ was the error term. β^ψ was estimated with signal strength incorporated, while the value of t_i was iteratively refined with the following equation:

where represented the current estimate of f(t). The centreline was calculated to minimize orthogonal projections of the data onto the fitted curve. The length of radionuclide distribution is then numerically calculated with:

The distribution of surrogates was occasionally discontinuous in which case the imaging data had to be split into parts. The fitting curves to each individual part were then combined, as shown in Figure 1D–F

Figure 1.

Spatial distribution of rectal microbicide tracked with SPECT/CT imaging. The SPECT/CT images were taken at 24 h after rectal dosing of ¹¹¹In-labelled microbicide surrogate. The SPECT ¹¹¹In signal is shown in blue; the co-registered CT anatomical signal is shown in grey. (A) Coronal view of an anterior anatomical plane. The microbicide signal can be seen in the rectosigmoid and descending colon. (B) Coronal view of a posterior plane. The microbicide signal can be seen at the splenic flexure, transverse, and ascending colon. (C) Coronal maximum intensity projection of SPECT signal. The box indicates the split of data into two parts for later 3-D curve fitting to each part. (D) 1000 random samples were taken from the data in the box shown in C. The blue-green-orange colour shows the signal intensity in increasing order. The two red dots indicate the start and end points of microbicide pathway based on visual inspection. (E) The centreline (red curve) was found through 3-D curve fitting. It passes through three constrained points (red dots). The orange line segments show the residuals. (F) The fitted curve superimposed on sampled points. (G) The concentration–distance curve showing the schematic location of three spatial kinetic parameters, D_max (most proximal extent of signal), D_Cmax (distance at maximum concentration) and D_ave (mean residence distance)

Reference point of the centerline

To summarize the spread of surrogate across different subjects and different imaging sessions, a reference point was needed. At first, the anal verge was identified in the MRI image and the distance from the anal verge to the start of the SPECT signal was estimated. However, the uncertainty of using the anal verge as reference start was high due to the reduced sensitivity of CT images to discriminate differences in soft tissue. Therefore, a coccygeal reference point was used instead. This coccygeal reference point was defined through the identification of a transverse plane crossing the coccyx in the SPECT/CT fusion image. Since CT images were co-registered with the SPECT image, the coordinates of the coccyx could be converted into the indices of SPECT array in R. We described the centreline start at (a, b, c) with a representing the coordinate from right to left, b from back to front, and c from head to foot. The coordinates of the coccyx from head to foot can be described as m. The point (a, b, m) was termed the coccygeal reference point for the calculation of spatial parameters. For reference, where it could be unambiguously imaged with MRI or CT in our studies, the coccygeal reference point relative to the superior aspect of the internal anal sphincter ranged from 1.3 to 1.9 cm in the vertical dimension and 5.4 to 6.2 cm in the anterior-posterior dimension.

Mass density of radiotracers along the centerline

After the centreline was identified, the total amount of signal density at a radius of ∼1 cm (125 voxels) from each point of the centreline was calculated as a concentration measurement.

Pharmacokinetic concentration–distance parameters

The concentration–distance curve was constructed as shown in Figure 1G. Three novel pharmacokinetic parameters were computed from such plots. Maximum distance (D_max) is the distance of the proximal endpoint of the centreline from reference start, describing the furthest point that the radiosignal was detected along the luminal path of the colon. Distance at concentration maximum (D_Cmax) is the location of the colon where the highest drug density (or concentration) appeared and is measured as the distance on centreline away from the reference point. Mean residence distance (D_ave) was defined as ∫xf(x)dx divided by ∫f(x)dx where x is the distance from reference point and f(x) is the mass density along the centerline.

MRI

MRI data were collected as previously described [8] with a 1.5 T GE Signa LX scanner (GE Medical Systems, Waukesha, WI) with a pelvic phased array coil with T2-weighted and T1-weighted Spin Echo sequences with fat saturation. D_max was estimated by summing a series of line segments along the recto-sigmoid arc (and beyond) starting from the most caudal to the most cephalad location performed on the Advantage Windows workstation (GE SUN Advantage Windows AW4.0_02, GE Medical Systems, Waukesha, WI).

Statistical analysis

The geometric mean ratio and 95% confidence interval was calculated for each pharmacokinetic parameter and parameter differences comparing across parts and phases. The non-parametric sign test was also used for comparisons across test conditions. The Spearman correlation test was used to explore relationships between imaging methods and endoscopy.

Results

Nine subjects, aged 25 to 57 years, enrolled. Five completed all study parts, one subject was never dosed, one subject only completed phase A and was replaced with a subject who only completed phase B, two subjects received only one dose and were not replaced. There were a total of 31 imaging sessions. Several of the 10 h and 24 h scans were not performed due to absence of signal in the colon. No serious adverse events were reported.

As visualized using SPECT/CT, gel distribution was limited to the recto-sigmoid colon in 84% (26 of 31) of imaging sessions. A distinctly different distribution pattern up into the descending colon was seen in 16% (5 of 31) of imaging sessions. These were all from three subjects, two of whom had more than one imaging session with this long proximal distribution. In the subject shown in Figure 2 (part 1, phase A), about 30 min after dosing, the microbicide surrogate was distributed within the recto-sigmoid colon. It moved into the distal descending colon 4 h later and continued moving cephalically along the descending colon. After 24 h, the vehicle reached the splenic flexure. Such broad distribution was also seen in Part III, Phase B in this subject, and in three imaging sessions in the two other subjects.

Figure 2.

The time course of spatial distribution of microbicide vehicles. This subject received ^99mTc-labelled gel, followed by simulated intercourse. The SPECT/CT images were taken at 0.5, 4, 10 and 24 h after dosing. (A) SPECT/CT fusion imaging. The microbicide surrogate is shown in blue. The co-registered CT image is in greyscale. In each panel, the coronal section is on the left, the sagittal section in the middle and maximum intensity projection (SPECT only) in the coronal plane on the right. * indicates radiotracer on body surface used for anatomical reference. (B) 1000 random samples taken from each SPECT image and plotted in three dimensions. The blue-green-orange colour shows the signal intensity in increasing order. The red curve shows the centreline found through 3-D curve fitting [green = left (sagittal); yellow = back (coronal); dark yellow = bottom (transverse)]. Scale changes with each scan to maximize detail. (C) Concentration–distance plot. (D) Spatial parameters vs. time plot. D_max (); D_ave (); D_Cmax ()

Two participants had longer distributions compared with the other participants and were useful to demonstrate curve fitting and PK parameter estimates. One subject had migration of signal retrograde to the ascending colon over 24 h, farther than in any other research participant (Figure 1A–C). This participant is presented as an illustration of the curve-fitting process (Figure 1D–F), concentration vs. distance plotting (Figure 1G), and spatial kinetic parameter estimation (D_max, D_Cmax, D_ave, Figure 1G) for a single SPECT/CT study. For the second participant, the dimension of time is added to summarize the 24 h sequence of SPECT/CT scans (Figure 2). The series of scans (Figure 2A) is processed using curve-fitting (Figure 2B), concentration–distance construction (Figure 2C) and plotting spatial kinetic estimates vs. time (Figure 2D) to provide a comprehensive summary of concentration–distance–time relations for a single dose over 24 h.

Comparison of D_max, D_Cmax, and D_ave between any two parts of the study did not reveal any statistically significant difference, except that the D_Cmax and D_ave were larger with concomitant microbicide and semen surrogates (part III) than with semen surrogate alone (part II) in phase A for all five subjects (Sign test, P = 0.03). After pooling phases A and B to increase the number of pairs for comparison, the only significant difference was the 4 h semen surrogate (part II) D_Cmax which was greater than the microbicide surrogate (part I) with geometric mean difference of 2.9 cm (95% CI 0.0, 5.8).

Overall, from the initial observation to the 4 h observation the surrogates moved cephalically about 5 cm in terms of D_max and about 2 to 3 cm in terms of D_Cmax and D_ave. The surrogates remained largely in place for the next 6 h (Table 2). At 24 h, the surrogates reached 18 cm (geometric mean of D_max) beyond the coccygeal plane, a retrograde migration of 8 cm from the initial SPECT/CT scan. The D_ave was also significantly larger at 24 h, indicating 3.4 cm retrograde migration beyond the initial observation while D_Cmax remained unchanged.

Table 2.

Summary of spatial pharmacokinetic parameters

	Dmax (cm)		DCmax (cm)		Dave (cm)
Post dosing (h)	Geometric mean (95% CI)	Median (Range)	Geometric mean (95% CI)	Median (Range)	Geometric mean (95% CI)	Median (Range)
0.5	10 (8.6, 12)	9.8 (5.8–21)	3.8 (2.7, 5.3)	4.4 (1.1–8.7)	4.2 (3.5, 5.1)	4.3 (1.9–7.2)
4	15 (13, 18)	14 (9.0–63)	6.0 (5.0, 7.2)	6.0 (2.0–14)	6.9 (5.8, 8.2)	6.7 (3.2–29)
10	14 (10, 18)	13 (7.3–38)	5.2 (3.8, 7.2)	5.1 (2.6–13)	5.6 (4.0, 7.8)	5.7 (2.4–15)
24	18 (13, 26)	16 (7.3–105)	4.2 (2.8, 6.3)	4.4 (0.52–15)	7.6 (5.3, 11)	7.2 (2.7–40)

For the selected subject shown earlier (Figure 1), the concentration–distance pattern appears somewhat dissimilar at both 4 and 24 h times of observation when comparing direct endoscopic brush sampling, which had a smaller number of concentration peaks, with indirect SPECT/CT imaging methods (Figure 3). Similar results were seen in two other image sessions (data not shown). Looking at aggregate concentration–distance data for part II (semen surrogate) studies to provide a global comparison of endoscopic vs. SPECT/CT methods (Figure 4), both modalities indicate peak concentrations at recto-sigmoid locations which decline as one moves cephalad towards the descending colon. Still, however, the methods do not demonstrate concurrent concentration–distance patterns as peak number and rate of change are different throughout the observed distances.

Figure 3.

Concentration–distance plot comparing indirect (SPECT/CT imaging) and direct (endoscopy) quantitative methods for the subject shown in Figure 1. The distance is relative to the coccygeal reference point for imaging data and anal verge for endoscopic brush data. Imaging (); Brush ()

Figure 4.

Concentration–distance plot comparing indirect (SPECT/CT imaging) and direct (endoscopy) quantitative methods for all subjects. The imaging data were pooled from data collected in study part II and smoothed with Loess function with sampling proportion of 0.1 and polynomial degree of 1. The brush data (collected following imaging in study part II) were averaged from all subjects and corrected for radioactive decay to allow comparison across time. The error bar indicates standard error (n = 14). The distance is relative to the coccygeal reference point for imaging data and anal verge for endoscopic brush data. Brush (); Image ()

In comparing SPECT and MRI, we selected D_max as a distance parameter suitable for comparison across both methods (Figure 5). We identified a statistically significant, but very low magnitude, correlation when pooling all SPECT-MRI pairs (Spearman correlation test, r = 0.29, P = 0.06, n = 41 pairs). When limiting the correlation to any specific time or removing the four highest SPECT/CT D_max pairs, none of these more limited sets was significant. Because we could not estimate a concentration equivalent in the MRI scans, comparison of the other distance parameters (D_Cmax and D_ave) were not possible.

Figure 5.

Comparison of maximal distance of signal migration as assessed with MRI and SPECT imaging. 0h (•); 4h (▿); 10h (); 24h (◊)

Discussion

In this study, we described the concentration along the distance of the centreline path at each sampling time and estimated three new spatial kinetic parameters to aid in the quantitative description of drugs and semen surrogates in the distal colon. The imaging method was non-invasive, allowed tracking the movement of drugs and HIV surrogates in the whole colon, and was not limited to a number of discrete locations acquired via invasive endoscopy sampling, which might also disturb the distribution process. MRI can provide indirect, unperturbed assessment of the continuous colon. In contrast to SPECT, however, MRI was hindered by difficulty distinguishing low gadolinium signal against background, a smaller field of view and an inability to provide an estimate of concentration. These limitations resulted in different and, we believe, less accurate results compared with SPECT/CT. We have subsequently applied these dual isotope methods to assess co-localization of cell-free and cell-associated HIV surrogates in both the rectum and vagina [16, 17].

The concentration–time profile can differ between blood plasma and the site of drug action. The classic example is the concentration–response relationship of d-tubocurarine which sparked the proposal of a hypothetical effect compartment [18]. Because the initial events of rectally acquired HIV infection are believed to be local and not systemic, we needed a means to describe the spatial distribution of both drug and virus in quantitative terms to enable evaluation of candidate drug distribution relative to viral distribution. The local environment of the colon is three-dimensional, making the extraction of key pharmacokinetic information more complicated. This difficulty was approached through the establishment of centreline and the concentration estimation along the centreline. Peaks, valleys, and discontinuities are apparent in the concentration–distance curves. This topology likely results from segmentation of colonic contents resulting from ongoing circumferential peristaltic activity that continues after the dose. The peak number increases with time (Figure 2C and 4). This is less evident with endoscopic sampling due perhaps to a gel spreading effect of the sigmoidoscope. To enable a quantitative summary of the spatial distribution, we defined three new distance-related pharmacokinetic parameters.

D_max indicates how far a pathogen in semen (e.g. HIV) spreads within the colon and whether a microbicide reaches far enough in comparison. D_Cmax indicates the location of highest concentration of a pathogen and, therefore, the location of interest for a target drug concentration in order to achieve adequate stoichiometry of drug and pathogen for efficacy. D_ave is similar to a mean residence time, but with distance, indicating the mean distance within the colon throughout which the pathogen or drug is distributed at the time of the observation. This is useful in the presence of multiple peaks which are commonly seen. These spatial parameters might be used in a way similar to traditional pharmacokinetic parameters (C_max and AUC) during the development of rectal microbicides, but with application to drugs for other diseases of the colon and, possibly by extension, to vaginal microbicide development. Unlike traditional pharmacokinetic parameters such as C_max and AUC which quantify drug at a given location across different sampling times, our spatial pharmacokinetic parameters quantify the drug at a given time across different locations. Since drug targets are often expressed in different regions, quantification of the difference in spatial distribution of drugs may help the development of other targeted therapies as well.

Overall, the concentration–distance curves obtained with SPECT/CT and sigmoidoscopy appeared roughly similar as observed previously with an earlier SPECT concentration–distance estimation method [9]. However, the details of the distance–concentration curves developed by SPECT and sigmoidoscopy are clearly different as may be expected. Sampling via endoscopic brush may create a measurement artifact by stimulating muscular action of the colon, distorting the colon path by stretching out natural circumferential colonic folds, and physically displacing contents of gel or other substances within the colon lumen. SPECT/CT imaging is non-invasive and provides millions of concentration data points without interfering with the colonic function, anatomy or contents, and allowing multiple sequential imaging. Coupled with our quantitative methods, SPECT imaging has a clear advantage over an endoscopic sampling method for the determination of luminal pharmacokinetics.

When compared with MRI, SPECT/CT has superior sensitivity and is more amenable to concentration estimates. When evaluating the distribution of gadolinium within the colon with MRI, a diluted gadolinium signal and the possibility of other proteinaceous material with similar bright signal intensity (not a limitation of SPECT) limited precision of the estimates. The more limited field of view of our MRI studies may have caused the only weak correlation of D_max estimates. For locations like the vagina, with smaller, less complex luminal anatomy and less confounding or dilution of gadolinium signal, however, MRI may be superior.

Our use of a sexual lubricant gel as a semen surrogate and sulphur colloid as a HIV surrogate differ from semen and HIV. Semen is quite a different, complex and dynamic protein and glycan mixture, sulphur colloid lacks the specific HIV–mucosal surface interactions and cell-associated HIV is not represented. We used autologous human semen and radiolabelled autologous lymphocytes in a separate study to overcome some of these limitations [16].

In summary, spatial kinetic parameters estimated following 3-D curve fitting enabled summarization of the movement of microbicide and HIV surrogates in the distal colon. The method should facilitate comparisons between formulation candidates to aid selection of optimal formulation and dose frequency to further rectal microbicide development. Combination of these methods to evaluate simultaneously a microbicide candidate and an HIV surrogate could guide evaluation of candidate microbicide distribution in light of the simultaneous distribution of the HIV it is intended to prevent.

Competing Interests

EJF has received salary support from the National Institutes of Health, the Foundation for AIDS Research, International Partnership for Microbicides and CONRAD. CWH has received financial support for research, managed by John Hopkins University, from Gilead Sciences (Foster City, CA).