Detection of prostate carcinoma with contrast-enhanced sonography using intermittent harmonic imaging

Authors

  • Ethan J. Halpern M.D.,

    Corresponding author
    1. Department of Radiology, Jefferson Prostate Diagnostic Center, Kimmel Cancer Center, Thomas Jefferson University, Philadelphia, Pennsylvania
    • Thomas Jefferson University, 132 South 10th St., Philadelphia, PA 19107-5244
    Search for more papers by this author
    • Fax: (215) 955-8549

  • John R. Ramey M.D.,

    1. Department of Urology, Jefferson Prostate Diagnostic Center, Kimmel Cancer Center, Thomas Jefferson University, Philadelphia, Pennsylvania
    Search for more papers by this author
  • Stephen E. Strup M.D.,

    1. Department of Urology, Jefferson Prostate Diagnostic Center, Kimmel Cancer Center, Thomas Jefferson University, Philadelphia, Pennsylvania
    Search for more papers by this author
  • Ferdinand Frauscher M.D.,

    1. Department of Radiology, Jefferson Prostate Diagnostic Center, Kimmel Cancer Center, Thomas Jefferson University, Philadelphia, Pennsylvania
    Search for more papers by this author
  • Peter McCue M.D.,

    1. Department of Pathology, Jefferson Prostate Diagnostic Center, Kimmel Cancer Center, Thomas Jefferson University, Philadelphia, Pennsylvania
    Search for more papers by this author
  • Leonard G. Gomella M.D.

    1. Department of Urology, Jefferson Prostate Diagnostic Center, Kimmel Cancer Center, Thomas Jefferson University, Philadelphia, Pennsylvania
    Search for more papers by this author

Abstract

BACKGROUND

The purpose of this study was to assess prostate carcinoma detection and discrimination of benign from malignant prostate tissue with contrast-enhanced ultrasonography.

METHODS

In all, 301 subjects referred for prostate biopsy were evaluated with contrast-enhanced sonography using continuous harmonic imaging (CHI) and intermittent harmonic imaging (IHI) with interscan delay times of 0.2, 0.5, 1.0, 2.0 seconds, as well as continuous color and power Doppler. Targeted biopsy cores were obtained from sites of greatest enhancement, followed by spatially distributed cores in a modified sextant distribution.

RESULTS

Carcinoma was detected in 363 biopsy cores from 104 of 301 subjects (35%). Carcinoma was found in 15.5% (175 of 1133) of targeted cores and 10.4% (188 of 1806) of sextant cores (P < 0.01). Among subjects with carcinoma, targeted cores were twice as likely to be positive (odds ratio [OR] = 2.0, P < 0.001). Clustered receiver operating characteristic (ROC) analysis of imaging findings at sextant biopsy sites yielded the following Az values: precontrast gray scale: 0.58; precontrast color Doppler: 0.53; precontrast power Doppler: 0.58; CHI: 0.62; IHI (0.2 sec): 0.64; IHI (0.5 sec): 0.63; IHI (1.0 sec): 0.65; IHI (2.0 sec): 0.61; contrast-enhanced color Doppler: 0.60; contrast-enhanced power Doppler: 0.62. A statistically significant benefit was found for IHI over baseline imaging (P < 0.05).

CONCLUSIONS

The carcinoma detection rate of contrast-enhanced targeted cores is significantly higher when compared with sextant cores. Contrast-enhanced transrectal sonography with IHI provides a statistically significant improvement in discrimination between benign and malignant biopsy sites. However, given the relatively low ROC areas, this technique may not be sufficient to predict which patients have benign versus malignant disease. Cancer 2005. © 2005 American Cancer Society.

The number of new cases of prostate carcinoma that will be diagnosed in the U.S. for 2005 is estimated at 232,090, with 30,350 deaths.1 Between 1986 and 1991 the rate of prostate needle biopsy in men older than 65 years of age increased from 685 to 2600 per 100,000.2 Because the proportion of prostate biopsies positive for carcinoma is slightly under one-third, the number of prostate biopsies performed annually in the U.S. in 2005 is estimated to be greater than 700,000.

The sextant biopsy protocol, a systematic, spatially distributed set of six biopsy cores obtained under transrectal ultrasound guidance, was described in 1989, and remained the standard of care for a decade.3 More recently, clinicians have begun to advocate 10–12 biopsy cores4, 5 or a ‘saturation biopsy’ approach.6 Such systematic techniques increase the volume of sampled tissue, but do not identify and target specific lesions. Because patient morbidity and pathology costs are related to the number of biopsy cores, a targeted biopsy approach that could maintain the efficiency of cancer detection with a reduced number of biopsy cores would represent a cost-effective approach to the diagnosis of prostate carcinoma.

Carcinoma of the prostate is classically described as hypoechoic,7 but can appear echogenic or isoechoic.8 Color Doppler imaging has been proposed to supplement conventional gray-scale imaging.9, 10 Increased color Doppler signal correlates positively with both prostate tumor stage and grade, as well as with the risk of recurrence after treatment.11 Power Doppler may be even more useful in the detection of prostate carcinoma.12, 13 Nonetheless, conventional color and power Doppler-guided needle biopsy do not substantially improve the detection rate of prostate carcinoma.14 The combination of gray scale and Doppler ultrasound is not sufficient to eliminate the need for systematic biopsy.15–17

Microbubble contrast agents enhance sonographic visualization of the microvasculature associated with prostate carcinoma.18, 19 These agents increase the echogenicity of the intravascular space on gray-scale harmonic imaging, and also provide a dramatic visible increase in Doppler signal. Intermittent imaging is an ultrasound technique that employs a reduced frame rate, allows more time for contrast agent to enter the scan plane between frames, and thereby increases the intensity of microbubble contrast enhancement.20–22 Preliminary data suggest that intermittent gray-scale harmonic imaging (IHI) can increase the conspicuity of microvascular enhancement associated with prostate carcinoma.23 The current study was designed to evaluate the ability of contrast-enhanced sonography with IHI to improve the detection of prostate carcinoma with a targeted biopsy technique.

METHODS

Study Population

Institutional review board approval was obtained for this Department of Defense-sponsored protocol, and written informed consent was obtained from each participant. Three hundred and one subjects with an elevated prostate-specific antigen (PSA ≥ 4 ng/mL) or abnormal digital rectal examination were enrolled between October 2001 and January 2004. Mean patient age was 63 ± 8 with a range of 42–87. Mean PSA was 9.5 ± 23 ng/mL with a range of 0.4–360.7. There were 73 subjects with a PSA below 4.0 ng/mL, 118 subjects with a PSA above 4 and below 10, 54 subjects with a PSA in the range of 10–35, and 6 subjects with a PSA above 35. The population consisted of 232 Caucasian males, 52 African American males, 3 Asian Indian males, 8 Asian males, 2 Philippine males, 3 Hispanic males, and 1 Asian/Hispanic male. Just under half of the study subjects (n = 134) had a previous negative biopsy procedure of the prostate with a PSA that remained elevated.

Imaging Protocol

Sonography was performed with the Sonoline Elegra system (Siemens Medical Systems, Issaquah, WA) using a 6.5 MHz end-fire transducer. To reduce the impact of patient position on prostatic blood flow, all subjects were examined in the lithotomy position.24 Gray-scale imaging was performed with a center probe frequency of 5.14 MHz, a dynamic range of 55 dB, and a persistence setting of 2. Continuous gray-scale harmonic imaging (CHI) was performed with a default mechanical index of 0.4. The mechanical index was automatically increased into the range of 0.8–1.0 for IHI. For color and power imaging, the center probe frequency was at 4.0 MHz with a dynamic range of 30 dB, pulse repetition frequency of 868 Hz, and wall filter set to low. The Doppler window for color and power imaging included the entire gland. Color and power gain were adjusted to maximize signal but eliminate color noise from the tissue of the prostate. The entire examination was recorded on sVHS videotape.

To obtain comparable images from the precontrast and postcontrast portions of the examination, multiple identical series of angled axial sweeps through the gland were obtained from base to apex, each sweep extending over a period of 20–30 seconds. Precontrast imaging sweeps were performed with conventional gray-scale imaging as well as gray-scale harmonic imaging, color Doppler, and power Doppler. Postcontrast imaging sweeps were performed with gray-scale harmonic imaging in continuous mode and repeated during IHI with interscan delay times of 0.2, 0.5, 1.0, and 2.0 seconds. Two additional postcontrast imaging sweeps were performed with continuous color Doppler and continuous power Doppler imaging.

Contrast Infusion

Contrast-enhanced imaging was performed during infusion of AF0150 (Imagent, formerly: Alliance Pharmaceutical, San Diego, CA; now: Imcor; San Diego, CA). AF0150 is a sterile, nonpyrogenic white to off-white powder of spray-dried microspheres. The microspheres consist of surfactants, buffers, salts, and a water-soluble structural agent that dissolves when reconstituted, forming a dispersion of stable and highly echogenic microbubbles (typical volume-weighted mean diameter of 6 μm) in a buffered, iso-osmotic solution. AFO150 remains within the circulation for several minutes after injection and produces both gray-scale and Doppler enhancement.25

On the basis of information gained in the previous trials of Imagent, contrast material was delivered by intravenous infusion during the prostate examination and biopsy procedure. A dose of 4.0 mg/kg of Imagent was added to 150 mL of normal saline. For the average patient this dose amounted to two patient contrast kits with a retail cost of $250 ($125 per patient kit). The initial infusion rate was 8 mL/min. Postcontrast imaging began as soon as the contrast was visible on continuous gray-scale imaging. The infusion rate was adjusted in the range of 8–12 mL/min to subjectively optimize visible enhancement of the prostate. The infusion continued for approximately 10 minutes, during which time imaging sweeps and biopsy were performed.

Biopsy Protocol

After the completion of contrast-enhanced imaging, ultrasound-guided biopsy was performed with an 18-gauge automated spring-loaded biopsy gun. Topical anesthesia was given with lidocaine gel, but a transrectal injection of lidocaine was not performed because of concerns related to possible effects of injected lidocaine on blood flow within the prostate. Up to four targeted-biopsy specimens per prostate (per patient) were first obtained from areas with the greatest amount of contrast enhancement, followed by a modified sextant biopsy. The modified sextant cores were obtained from the areas of greatest flow in the outer gland at the base, midportion, and apex, on both the right and left sides. These cores were often laterally directed and often overlapped with the directed cores. When an area of increased flow was not identified within a particular sextant, a laterally directed core was obtained for the modified sextant protocol. Thus, six modified sextant cores were obtained from each subject along with 0–4 targeted cores.

Image Interpretation

A subjective rating score was assigned for each sextant biopsy site on each imaging sequence. Rating scores were assigned by the examining physician at the time of the initial examination. An independent interpretation of the imaging findings was performed by a second physician who reviewed a videotape of the examination, but was blinded to the initial interpretation as well as to all clinical and pathological information. The examining physician was an experienced radiologist who had performed many previous studies with contrast-enhanced ultrasound of the prostate. An experienced urologist who had performed many prostate biopsy procedures, but had not previously worked with contrast-enhanced ultrasound, performed the independent, blinded interpretations for the first 100 subjects. A junior urologist with little prior biopsy experience trained together with the examining physician during performance of the first 100 studies. This junior urologist performed the remaining independent blinded interpretations.

For baseline gray-scale imaging, the suspicion of carcinoma was scored based on tissue echotexture and gland contour abnormalities. For baseline color and power Doppler the score was based on the level of Doppler flow observed. For postcontrast imaging, each sextant biopsy site was scored for the level of contrast enhancement. A five-point subjective scale was used with the following general guidelines.

Baseline Gray-Scale Scoring System

  • 1Normal appearance (homogeneous, echogenic outer gland).
  • 2Probably normal (minimal heterogeneity of the outer gland).
  • 3Indeterminate (contour asymmetry or ill-defined echotexture abnormality).
  • 4Probably carcinoma (focal contour bulge or probable mass).
  • 5Definitely carcinoma (focal hypoechoic mass).

Baseline Doppler Scoring System

  • 1Normal appearance (capsular and periurethral flow only).
  • 2Probably normal (symmetric radial flow extending in from capsular branches).
  • 3Indeterminate (subtle asymmetric/increased flow pattern in outer gland).
  • 4Probably carcinoma (definite asymmetric/increased flow in outer gland).
  • 5Definitely carcinoma (focal asymmetric/increased flow with disorganized pattern).

Contrast-Enhanced Scoring System

  • 1Minimal enhancement (capsular and periurethral flow only).
  • 2Mild enhancement (symmetric radial flow from capsular branches).
  • 3Mildly increased enhancement (asymmetric/increased flow in prostate).
  • 4Moderately increased enhancement (asymmetric/increased flow in prostate).
  • 5Substantially increased enhancement (asymmetric/increased flow in prostate).

Analysis

The number of carcinomas found and the percentage of positive cores were tabulated for both the targeted biopsy cores and the modified sextant cores (which are also targeted within each sextant). These results were further stratified by the number of previous biopsy procedures.

To compare the “by-core” positive biopsy yield of the targeted technique to the positive yield of the modified sextant technique, conditional logistic regression analysis was performed. Conditional logistic analysis was chosen because of clustered (nonindependent) biopsy sites within each subject. An odds ratio (OR) and corresponding confidence interval (CI) were computed for the detection of carcinoma in targeted versus sextant biopsy cores (STATA 8; Stata, College Station, TX).

To compare the ‘by-patient’ detection rate of carcinoma for targeted and sextant biopsy techniques, carcinoma detection with targeted and sextant techniques was tabulated by patient. A McNemar's chi-square was computed to compare the carcinoma detection rate with the two techniques (STATA 8). To determine whether there was a difference in location of carcinoma detected by targeted versus sextant biopsy, the number of positive cores obtained from the base, midgland, and apex was tabulated, and a chi-square test for trend was performed (Epi-Info 6; Centers for Disease Control and Prevention, Atlanta, GA). To determine whether the targeted technique detected higher-grade carcinomas, the Gleason scores of carcinomas detected by targeted and sextant biopsy were tabulated and compared with a chi-square test for trend (Epi-Info 6).

Receiver operating characteristic (ROC) analysis was used to assess the diagnostic accuracy of ultrasound for the detection of prostate carcinoma based on the 5-point subjective scores at each sextant biopsy site. To compensate for the lack of independence among the sextant biopsy sites within an individual patient, clustered ROC analysis was performed.26 To avoid the bias that would result from redundant biopsies at sites with greater enhancement, only the sextant biopsy data was used for the clustered ROC analysis.

To determine whether contrast-enhanced ultrasound and targeted biopsy would selectively detect additional forms of prostate pathology, biopsy cores with a finding of prostatic intraepithelial neoplasia (PIN), atypical small acinar proliferation (ASAP), and prostatitis were identified. Conditional logistic regression was used to determine whether there was a statistically significant increased probability of detecting these three types of prostate pathology with targeted biopsy.

To evaluate interobserver agreement, a kappa score was computed to quantify interobserver agreement between the rating scores of the examining physician and each of the two blinded readers. A quadratic weighted kappa was used to accommodate the 5-point rating score for each biopsy site (STATA 8).

RESULTS

Infusion of contrast material resulted in visible vascular enhancement in every subject. Total examination time for the baseline and contrast-enhanced study was extended by about 15 minutes compared with the time required for a noncontrast study. One patient experienced a delayed allergic reaction that may have been related to antibiotic prophylaxis. Another patient experienced a severe vasovagal episode that appeared to be related to the biopsy procedure. No adverse events related to the contrast agent were observed.

Illustrations of contrast-enhanced imaging of the prostate are presented in Figures 1 and 2. Figure 1 demonstrates a hypoechoic Gleason 8 carcinoma on baseline imaging that demonstrates contrast enhancement. Figure 2 demonstrates another Gleason 8 carcinoma that is not clearly defined with baseline gray-scale or Doppler imaging, but is clearly enhanced on postcontrast gray-scale harmonic and Doppler imaging. The tumor blush appears qualitatively different from IHI as compared with CHI.

Figure 1.

A 75-year-old male with Gleason 8 carcinoma in the left mid-gland. (A) Baseline color Doppler image demonstrates a hypoechoic mass in the left mid-gland. (B) Postcontrast CHI demonstrates enhancement of the carcinoma (arrow). (C) Postcontrast IHI with 1.0-second interscan delay demonstrates larger blush of tumor enhancement (arrow). (D) Postcontrast color Doppler demonstrates enhancement of tumor. (E) Postcontrast power Doppler demonstrates enhancement of tumor.

Figure 2.

A 78-year-old male with Gleason 8 carcinoma in the left mid-gland. (A) Baseline color Doppler image demonstrates a small focal calcification in the left mid-gland, but not other evidence of mass or carcinoma. (B) Postcontrast CHI demonstrates enhancement of the carcinoma (arrow). (C) Postcontrast IHI with 2.0-second interscan delay demonstrates larger blush of subtle tumor enhancement (arrow). (D) Postcontrast color Doppler demonstrates enhancement of tumor.(E) Postcontrast power Doppler demonstrates enhancement of tumor.

Carcinoma was detected in 363 biopsy cores from 104 of 301 (35%) subjects. The positive biopsy rate is tabulated as a function of the number of previous biopsy procedures in Table 1. Carcinoma was found in 15.5% (175 of 1133) of targeted cores and 10.4% (188 of 1806) of sextant cores. Among subjects with carcinoma, targeted cores were twice as likely to return a positive biopsy (logistic regression OR = 2.0, P < 0.001).

Table 1. Positive Biopsy Rate in Study Patients as a Function of Number of Previous Biopsy Procedures
No. of previous biopsy proceduresNo. of patientsNo. of cancers% Positive biopsy procedures
01676841
1632235
242819
316319
≥ 413323

The diagnosis of carcinoma was discovered in 72 subjects by both targeted and sextant techniques. In 21 subjects carcinoma was detected by sextant biopsy alone and in 11 subjects carcinoma was detected by targeted biopsy alone (Table 2). A “by-patient” McNemar analysis demonstrated no statistically significant advantage of the targeted versus the sextant approach (P = 0.08).

Table 2. Cancers Detected by Sextant and Targeted Biopsy Approaches
 Targeted +Targeted −
Sextant +7221
Sextant −11197

The 21 subjects with carcinoma detected by sextant biopsy alone included 5 carcinomas at the gland base, 7 in the midgland, and 17 in the apex. The 11 subjects with carcinoma detected by targeted biopsy alone included 8 carcinomas at the gland base, 4 in the midgland, and 3 in the apex. While 38% (72 of 188) of positive sextant cores were obtained at the gland apex, only 17% (30 of 175) of positive targeted cores were obtained from the gland apex. Chi-square for trend analysis confirmed a statistically significant trend to find more carcinomas at the gland base with targeted cores and more carcinomas at the gland apex with systematic sextant cores (P = 0.006).

The distribution of Gleason scores in targeted and sextant cores is summarized in Table 3. Chi-square for trend analysis failed to demonstrate a significant relationship between biopsy technique and Gleason score (P = 0.36).

Table 3. Gleason Score Distribution as a Function of Biopsy Approach
 Gleason 5Gleason 6Gleason 7Gleason 8Gleason 9Gleason 10
Sextant alone1153110
Targeted alone101
Total257261432

To evaluate the ability of contrast-enhanced ultrasound to discriminate between benign and malignant tissue, clustered ROC analysis of sextant biopsy specimens was performed (Table 4). Carcinoma was detected in 188 sextant cores from 93 of 301 (31%) subjects. ROC areas for baseline gray scale and Doppler imaging ranged from Az = 0.53–0.58. ROC areas for contrast-enhanced imaging ranged from Az = 0.60–0.65.

Table 4. Areas Under the Curve (Az) for Clustered ROC Analysis Using Different Ultrasound Imaging Techniques
Ultrasound imaging techniqueArea under the ROC curve (Az)
Precontrast gray scale (baseline)0.58
Precontrast color Doppler (baseline)0.53
Precontrast power Doppler (baseline)0.58
Postcontrast gray scale harmonic imaging0.62
Intermittent gray scale harmonc imaging (0.2 sec)0.64
Intermittent gray scale harmonc imaging (0.5 sec)0.63
Intermittent gray scale harmonc imaging (1.0 sec)0.65
Intermittent gray scale harmonc imaging (2.0 sec)0.61
Contrast-enhanced color Doppler0.60
Contrast enhanced power Doppler0.62

A statistically significant benefit was found for all methods of postcontrast IHI over baseline gray-scale and Doppler imaging (P < 0.05). No significant difference in ROC area was observed with contrast-enhanced imaging at different interscan delay times. No single intermittent delay time was significantly superior for characterization of malignant sites. Furthermore, there was no statistically significant advantage for intermittent contrast-enhanced imaging beyond that provided by CHI. Although there was a statistically significant improvement in the characterization of tissue as benign versus malignant with contrast-enhanced imaging, the relatively low ROC areas (< 0.65) suggest that contrast-enhanced sonography did not definitively differentiate benign from malignant tissue without biopsy confirmation.

The diagnosis of prostatitis was made in 352 biopsy cores, including 158 of 1133 (13.9%) of targeted cores and 194 of 1806 (10.7%) of sextant cores (P = 0.72). The diagnosis of PIN was suggested in 52 of 1133 (4.6%) of targeted cores and 63 of 1806 (3.5%) of sextant cores (P = 0.70). The diagnosis of ASAP was suggested in 31 of 1133 (2.7%) of targeted cores and 47 of 1806 (2.6%) of sextant cores (P = 0.79). In contrast to the significantly increased detection of prostate carcinoma with targeted biopsy cores (OR = 2.0, P < 0.001), there was no statistically significant increase in the detection of prostatitis, PIN, or ASAP with targeted biopsy.

Interobserver agreement for the independent assessments of ultrasound findings are reported in Table 5. The value of kappa (κ) ranges from 0 (no agreement) to 1 (perfect agreement). κ was tabulated separately for the first blinded reader (experienced urologist without prior experience using contrast agents) and the second blinded reader (junior urologist with several months of training in contrast-enhanced imaging). The kappa values in Table 5 demonstrate better interobserver agreement with the second blinded reader, with the best agreement using enhanced Doppler imaging (κ = 0.54–0.55). On rereview of discrepant cases some of the differences between readers were related to subjective differences in evaluation of the level of enhancement. Another important source of interobserver discrepancy was disagreement as to location of enhancing areas within the prostate. When reviewing the study on videotape, the blinded reviewer was less certain of the image location than the primary examining physician.

Table 5. Interobserver Agreement between the Examining Physician and a Blinded Reader as Expressed by a Quadratic Kappa
Ultrasound imaging techniqueKappa: examining physician vs. reader no. 1Kappa: examining physician vs. reader no. 2
Precontrast gray scale (baseline)0.190.34
Precontrast color Doppler (baseline)0.280.46
Precontrast power Doppler (baseline)0.280.50
Postcontrast gray scale harmonic imaging0.170.53
Intermittent gray scale harmonic imaging (0.2 sec)0.180.47
Intermittent gray scale harmonic imaging (0.5 sec)0.190.49
Intermittent gray scale harmonic imaging (1.0 sec)0.160.46
Intermittent gray scale harmonic imaging (2.0 sec)0.070.35
Contrast-enhanced color Doppler0.120.54
Contrast enhanced power Doppler0.090.55

DISCUSSION

Persons with rising PSA are often subjected to multiple biopsy procedures. The positive biopsy rate on repeat sextant biopsy is approximately 19% after one initial negative biopsy, and drops to 8% after two negative biopsy procedures.27 The current study demonstrates that a targeted biopsy approach based on contrast-enhanced sonography can improve the detection of prostate carcinoma relative to sextant biopsy (OR = 2.0, P < 0.001). Among patients with a previously negative biopsy, carcinoma was detected in 22 of 63 (35%: 95% CI: 23–48%) of subjects with one prior biopsy and in 14 of 71 (20%: 95% CI: 11–31%) of subjects with multiple prior biopsies. Thus, the positive biopsy yield with our contrast-enhanced targeted biopsy is superior to the positive biopsy yield reported in the literature for repeat sextant biopsy.

Studies of microvessel density within the prostate demonstrate a clear association of increased microvessel density with the presence of carcinoma,28 with metastases,29 with the stage of disease,30–32 and disease-specific survival.33, 34 Quantitative assessment of microvascular density may actually provide important data to guide therapeutic decisions.35 However, the microvessels which proliferate in prostate carcinoma (10–30 μm) are below the resolution of conventional transrectal ultrasound. Microbubble ultrasound contrast agents represent one approach to observe these microvessels. Recently developed ultrasound contrast agents have intravascular residence times of several minutes, pass through the pulmonary circulation, and may be used for parenchymal organ enhancement.36, 37 Recent clinical trials have demonstrated improved detection of prostate carcinoma with targeted biopsy based on microbubble contrast agents.38–40

To enhance neovessels, contrast agents must pass into the microvascular circulation. Conventional gray-scale and Doppler imaging destroy most contrast microbubbles before they reach the microvasculature. However, IHI provides an interscan period during which contrast material may traverse further into the capillary bed without being destroyed.20–22 When compared with CHI, IHI provides a qualitatively different enhancement pattern based on penetration of contrast agent into smaller vessels, with improved contrast enhancement of prostate carcinoma.23 Our clustered ROC analysis confirm a statistically significant advantage of IHI over baseline imaging for the identification of prostate carcinoma, but little advantage for IHI over CHI (Table 4).

Although clustered ROC analysis demonstrates a statistically significant improvement in the discrimination of prostate carcinoma with contrast-enhanced imaging, the areas under the ROC curve (Az = 0.60–0.65) are only mildly superior to random chance (Az = 0.50). The relatively low ROC Az values for ultrasound detection of prostate carcinoma are not surprising in light of the poor interobserver agreement demonstrated in Table 5. Improved κ values for the second blinded reader (κ = 0.35–0.55) demonstrate the importance of training for interpretation of contrast-enhanced imaging.

Why is it so difficult to define prostate carcinoma with contrast-enhanced imaging? In contrast to the solitary, well-defined spherical tumors present in many solid organs, prostate carcinoma is multifocal in 85% of cases, and the individual sites of tumor are often oblong and irregular in shape.41 Prostate carcinoma often grows along the capsule of the prostate.42 Furthermore, the normal radial vascular pattern of the prostate is often distorted by the presence of benign prostatic hyperplasia. For these reasons, the hypervascularity associated with prostate carcinoma may not present as a round mass, and may be difficult to differentiate from normal capsular vascularity.

If the discriminatory ability of contrast-enhanced ultrasound for prostate carcinoma is poor (ROC Az < 0.65), why do targeted biopsy specimens double the positive yield for detection of cancer? As demonstrated in Figures 1 and 2, the higher positive biopsy yield of the targeted cores in this study confirms that areas of increased enhancement are more likely to contain a malignancy. Conversely, the relatively low ROC areas (Table 4) suggest that contrast-enhanced imaging cannot adequately discriminate benign from malignant areas. The explanation is related to the many false-positive sites of enhancement that reduce the specificity of contrast-enhanced imaging.43 To improve the discrimination of malignant from benign tissue, future efforts must concentrate on eliminating these areas of false-positive enhancement.

Although targeted biopsy detected 11% (11 of 104) of carcinomas not found by the sextant approach, targeted biopsy failed to detect 20% (21 of 104) of carcinomas. Among 21 subjects whose carcinoma was not detected by targeted biopsy, 17 subjects had tumor in the gland apex on sextant biopsy. Chi-square analysis for trend demonstrates a significant decrease in carcinoma detection by targeted cores toward the apex of the prostate (P = 0.006). To maximize carcinoma detection and minimize the number of biopsy cores, we suggest that a contrast-enhanced targeted biopsy strategy combined with additional “systematic” cores at the apex of the prostate be implemented.

Study Limitations

ROC analysis was applied to evaluate our ability to discriminate benign from malignant prostatic tissue. The original plan was to use a consensus ultrasound rating of the examining physician and the blinded reader for this ROC analysis. However, during consensus readings it was often difficult to precisely demonstrate the spatial correspondence between the diagnostic portion of the study and the biopsy sites. It became obvious that the original readings of the primary examining physician at the time of the prostate biopsy procedure corresponded more closely to the selected biopsy sites. On the basis of these considerations, we used the ultrasound interpretation provided by the examining physician for ROC analysis of contrast-enhanced ultrasound in the detection of prostate carcinoma.

The targeting of sextant cores to the most enhancing site within each sextant implies that the sextant cores in this study are not equivalent to a standard systematic sextant. A true comparison of targeted biopsy to the systematic sextant would require two independent examining physicians, one to perform the targeted biopsy and a second to perform the systematic sextant biopsy. The targeted sextant methodology selected for the current study would tend to increase the overlap between targeted and sextant biopsy cores, and should bias the comparison of targeted versus sextant biopsy toward a null result. Thus, our result for the comparison of targeted and sextant biopsy (OR = 2, P < 0.001) almost certainly underestimates the significant advantage of the targeted approach over a systematic sextant biopsy approach.

In actuality, the most vascular area within each sextant was almost always along the lateral margin of the prostate. Thus, the targeted sextant specimens were generally distributed in a similar pattern as the laterally directed sextant specimens. Nonetheless, given the potential sources of bias described above, the true difference in biopsy yield between the targeted approach and a systematic sextant approach is likely to be even larger than the difference reported in this study.

Why did we use a targeted sextant approach rather than a standard systematic sextant biopsy? This study was designed to evaluate both the detection of prostate carcinoma and the discrimination of benign from malignant prostate tissue with contrast-enhanced imaging. To evaluate the utility of contrast-enhanced imaging, it was critical that the biopsy specimens should be taken from the sites of maximum enhancement. However, ROC analysis based on targeted cores obtained at sites of increased enhancement would be biased by increased carcinoma detection secondary to multiple biopsy cores at enhancing sites. To avoid this potential bias, a targeted sextant approach was chosen with a single biopsy core corresponding to a single observer rating in exactly six locations in every prostate.

Pathology evaluation in this study is limited by the lack of correlation to whole-mount prostatectomy specimens. Additional sites of malignancy within the prostate may not be detected by needle biopsy. It is likely that the diagnosis of carcinoma is missed in a minority of patients by needle biopsy.44–46 Nonetheless, because the primary goal of the study was to evaluate the detection of prostate carcinoma, we chose to correlate imaging findings with the needle biopsy cores that are used for carcinoma detection. Furthermore, among the 301 subjects enrolled in this study, carcinoma was detected in 104. Because approximately 30% of patients diagnosed with carcinoma at our institution are treated with radical prostatectomy, the total number of patients with whole-mount correlation would be approximately 31. Whole-mount correlation for this subselected population would not provide information about carcinoma detection in the remaining 90% of our study patients.

We did not use transrectal injection of lidocaine in this study because of a fear that such an injection might alter the distribution of contrast enhancement in the prostate. Subsequent preliminary investigations in our department suggest that periprostatic injection of lidocaine does not visibly alter Doppler detection of blood flow around the prostate. However, based on the current study we cannot comment on the use of transrectal anesthetic in patients who are evaluated with contrast-enhanced sonography.

Conclusions

Intravenous infusion of a microbubble contrast agent provides sonographically visible enhancement of prostatic parenchyma, and can be used to target a biopsy procedure into areas of increased vascular flow. Among subjects with carcinoma, targeted cores were twice as likely to return a positive biopsy as compared with sextant cores (OR = 2.0, P < 0.001). With respect to the characterization of tissue as benign versus malignant, a statistically significant benefit was found for all methods of postcontrast IHI over baseline gray-scale and Doppler imaging (P < 0.05). However, there was minimal advantage to IHI beyond that provided by CHI or Doppler imaging.

Targeted biopsy of the prostate based on contrast-enhanced imaging does identify carcinomas that are not detected by conventional sextant biopsy. To maximize carcinoma detection with a minimum number of biopsy cores, we recommend a contrast-enhanced targeted biopsy strategy with additional systematic cores distributed to the apex of the prostate.

Acknowledgements

The authors thank the study coordinator, Ms. Dara DelCollo, for indispensable assistance.

Ancillary