Correlation of multiparametric MRI with histopathological... : Journal of Cancer Research and Therapeutics (2024)

INTRODUCTION

Prostatic cancer is the second most common malignant tumor in men, after lung cancer, with an annual incidence of 14.5%. Digital rectal examination (DRE) and serum prostate specific antigen (PSA) estimations are the commonly used screening tools. Transrectal ultrasonography (TRUS) has a higher sensitivity to detect small prostatic lesions that are negative on DRE.^[1]

Multiparametric MRI (Mp-MRI) helps in the detection of prostate cancer and the assessment of tumor aggressiveness based on the properties of malignant transformation such as increased cellular density detected on diffusion weighted imaging (DWI) with reduced apparent diffusion coefficient (ADC), increased neo-angiogenesis assessed by dynamic contrast enhanced MRI (DCE-MRI), and changes in concentration of intrinsic prostate metabolites such as increased choline and reduced citrate determined by magnetic resonance spectroscopy (MRS). Though Mp-MRI has emerged as a preferred diagnostic modality for imaging of prostate cancer, very few studies have been done to assess its utility in evaluating the aggressiveness of prostate cancer.^[2-6] This study attempts to determine the accuracy of DWI, DCE-MRI, and MRS for the detection of prostate cancer considering histopathology as its gold standard.

MATERIALS AND METHODS

The present study is a double blinded cross-sectional study, done in a tertiary care teaching institute from 1 April 2017 to 31 March 2019 for a period of 2 years after obtaining ethical approval. Male patients with lower urinary tract symptoms (LUTS) (i.e., frequency of urination, urgency, poor stream, prolonged micturition, feeling of incomplete bladder emptying, etc.), PSA value >4 ng/dL, and/or abnormal DRE findings were included in the study. Those previously diagnosed with acute prostatitis, diagnosed with prostate cancer, recent prostate biopsy, abnormal renal function tests, deranged coagulation profile or patients incompatible for MRI were excluded.

Of these, 44 patients with suspicious DRE findings (nodule, irregularity, hard consistency, obliteration of median sulcus) and elevated PSA underwent MRI on 1.5 Tesla MRI machine Magnetom Aera (Siemens, Germany) with the following protocol.

Axial T2WI was done at (TR-3000, TE-102, NEX 3), sagittal T2WI at (TR-7500, TE-101, NEX 3), turbo inversion recovery magnitude (TIRM) axial and coronal at (TR: 5241 ms, TE: 54 ms, NEX: 1 Matrix: 384, TI-160 ms) and T1 axial fat saturated at (TR: 612 ms, TE: 9.7 ms, NEX: 3). DWI (TR: 5100 ms, TE: 68 ms, Matrix: 192) was done using multiple b-values (50, 400, 800, 1000, and 1400 sec/mm²). DCE-MRI was done using rapid T1 weighted gradient echo imaging before, during, and after the intravenous administration of 8–10 mL gadopentetate dimeglumine (Magnilek) at 2–3 cc/sec starting with continuous image data acquisition. Post contrast volume interpolated breath-hold exam (VIBE): (TR: 4.20, TE: 1.58 ms, NEX: 1, Matrix: 192) T1 axial fat saturated post contrast (Turbo Spin Echo) (TR: 612 ms, TE: 9.7 ms, NEX: 3, Matrix: 256). All the slices were 3 mm thick with 0.3 mm interslice gap. Multi-voxel three-dimensional MRS using chemical shift imaging protocol was performed (TR: 930, TE: 120 ms, TA: 11:14, flip angle: 90°, bandwidth: 40 Hz, with water and lipid suppression).

The individual MRI parameters (T2, DWI, DCE-MRI, and MRS) were interpreted. Scoring was done from score 1 to score 5 on T2WI and DWI according to PI-RADS version 2 assessment category.^[7] Details of the scoring system are given in Table 1. ADC values were derived from DWI hyperintense areas that were hypointense on ADC map and which corresponded to hypointensity on T2WI. Quantitative values of ADC were obtained after placing region of interest (ROI) over the area with maximum hypointensity on the ADC map. Quantitative values of MRS were also noted. Both ADC and MRS values were also analyzed to correlate with grade of tumor. Choline to citrate (Cho/Cit) ratio was measured in suspicious areas to assess prostate cancer. The following metabolites were measured on ppm scale: citrate (2.6 ppm), creatine (3.0 ppm), choline (3.2 ppm).

Signal intensity versus time curve were obtained at the selected ROI and compared with the normal tissue. Three types of curves were obtained: Type 1 – progressive enhancement of the lesion, Type 2 – early steep slope of enhancement followed by plateau, Type 3 – early steep slope of enhancement followed by rapid washout of the contrast from the lesion.

Cognitive MR-TRUS fusion extended core prostate biopsy was carried out in all patients. Additional cores were taken from the suspected site of lesion apart from the 12 standard cores and were sent for histopathology. The pathologists were not aware of the MRI findings. Based on the Gleason score prostate cancer was graded either as high grade (Gleason score ≥7, excluding score 3 and 4) or low grade (Gleason score <7, including score 3 and 4).

The data were analyzed using SPSS version 20.0. Sensitivity, and specificity were calculated for T2WI, DWI, and ADC separately and combinedly, and PIRADS version 2 for prostate cancer. ROC curves were used to assess the area under the curve for ADC and Cho/Cit ratio and the best cut off for parameters for low-grade prostate cancer and high-grade prostate cancer was decided using the Youden index.

RESULT

A total of 52 lesions were noted in 44 patients –38 in the peripheral zone and 14 in the transitional zone. Only the peripheral zone lesions (38 lesions) were analyzed, for want of small number of prostate cancers detected in the transitional zone and imaging characteristic of prostatic lesion in these zones vary on imaging.

The mean age of the patients was 68.7 ± 10.1 (range 45–88) years. DRE findings were abnormal in all patients. The mean serum PSA level was 58.1 ± 22.4 ng/dL (range 4.5 to 108.3). Of the 38 indexed lesions in the peripheral zone, 33 (87%) had histopathologically proven to be prostate cancer. Out of which, 25 (65.7%) had high-grade prostate cancer and 8 (21.4%) had low-grade prostate cancer and 4 (10.5%) had non-specific chronic granulomatous prostatitis. No histopathological features of malignancy or prostatitis were noted in one case (2.6%); however, presence of fibrotic changes suggested an old fibrotic scar.

On qualitative scoring of T2WI, no lesion had score 1, three lesions had score 2, nine had score 3, seventeen had score 4, and nine had score 5. T2WI has a sensitivity and specificity of 75.8% and 80%, respectively, for diagnosis of prostate cancer (considering score 4 as cut off). On DWI, no lesion had score 1, one lesion had score 2, six had score 3, nineteen had score 4, and twelve had score 5, details of which has been given in Table 1. DWI has a sensitivity and a specificity of 90.9% and 80%, respectively, for detection of malignant prostatic lesion.

On qualitative DCE-MRI, out of 38 lesions, 10 lesions had type 1 curve; 12 lesions had type 2 curve, and 16 lesions had type 3 curve. On histopathology, 9 lesions (90%) with type 1 curve, 9 (75%) with type 2 curve, and 15 (93.7%) with type 3 curve, had prostate cancer. The results from the analysis showed that a rapid washout curve (Type 3) had lower sensitivity (45.5%) but high specificity (80%) for diagnosing prostate cancer.

Using prostate imaging – reporting and data system (PIRADS) version 2, on MRI 12 lesions were PIRADS-5, 20 were PIRADS-4, 5 were PIRADS-3, and 1 was PIRADS-2. Twelve (100%) of PIRADS-5, 19 (95%) with PIRADS-4; 3 (60%) with PIRADS-3, and no PIRADS-2 lesions had prostate cancer. PIRADS version 2 had a sensitivity and a specificity of 97% and 80.0%, respectively, considering PIRADS 4 as cut off. Sensitivity and specificity of various sequences have been given in Table 2.

The mean ADC value was significantly low in high-grade prostate carcinoma compared to that of low-grade carcinoma (0.675 ± 0.079 × 10⁻³ vs. 0.785 ± 0.095 × 10⁻³ mm²/sec, P = 0.008). The mean ADC value for prostate cancer (including low and high grades) was also significantly low (0.702 ± 0.094 × 10⁻³ mm²/sec), compared to prostatitis (0.959 ± 0.171 × 10⁻³ mm²/sec), P value 0.003 and normal prostatic parenchyma 1.31 ± 0.223 mm²/sec, P value = 0.001.

Prostate carcinoma can be differentiated from prostatitis if the lesion has an ADC below the cutoff value of 0.841 × 10⁻³mm²/sec with 75% sensitivity and 94% specificity (AUC = 0.962, P value 0.003). High-grade prostate carcinoma can be distinguished from low-grade prostate carcinoma if the ADC value is <0.752 × 10⁻³ mm²/sec with 84% sensitivity and 50% specificity (AUC = 0.815, P = 0.008) [Figure 1].

The mean Cho/Cit ratio for high-grade prostate cancer, low-grade prostate cancer and prostatitis were 2.60 ± 2.06, 1.5 ± 1.92, and 0.25 ± 0.50, respectively. The ROC curve of Cho/Cit ratio for differentiating prostate cancer from prostatitis has a cutoff value of 0.50 with a sensitivity and a specificity of 87.9% and 80.0%, respectively (AUC = 0.897, P value = 0.005) and for differentiating high-grade prostate cancer from low-grade prostate cancer, the cutoff value is 1.50 with a sensitivity and a specificity of 68.0% and 84.6%, respectively (AUC = 0.797, P = 0.003).

T2 hypointense lesions in peripheral zone showing significant diffusion restriction on DWI, type 3 curves in DCE-MRI and choline peaks in MRS were consistent with high-grade prostate cancer [Figure 2]. Low-grade prostate cancer were hypointense on T2WI, showing mild diffusion restriction and mildly increased choline [Figure 3]. Prostatitis lesions were T2 hypointense, not showing significant diffusion restriction or raised choline peak on MRS [Figure 4].

Other features of carcinoma included periprostatic spread in 14, seminal vesicle invasion in 14, urinary bladder wall in 4 and rectal wall invasion in 4, enlarged regional lymph node in 6 and distant metastasis in 4 patients.

DISCUSSION

Prostate cancer is most commonly seen in 60-years-old men who present with urinary symptoms, such as frequency of urination, urgency, poor stream, prolonged micturition, etc., and elevated PSA levels. In this study, we found prostatic carcinoma in the peripheral zone in 86.8% of cases. Most of the prostate cancer cases are noted in the peripheral zone as compared to the transition zone (8%).^[8]

Imaging methods, such as TRUS, colour Doppler, strain and shear wave elastography, contrast-enhanced Ultrasound (CEUS), have low sensitivity and specificity for detection of prostate cancer. Mp-MRI differs from conventional imaging as it includes a multitude of advanced sequences like DCE-MRI, DWI, and ADC maps and MRS in addition to conventional T1 and T2WI.

Zonal anatomy is well delineated on T2WI. The normal peripheral zone appears hyperintense due to its increased water content. Prostate cancer in the peripheral zone appears as an ill-defined, rounded focus of low signal intensity. However, other lesions such as prostatitis and fibrous scar may present as a similar hypointense lesion.^[9] T2WI plays an important role in local staging. Lesion extension into the seminal vesicle, obliteration of recto-prostatic angle, and neurovascular bundle represent extracapsular spread of tumor.

In this study, we found T2WI as the most sensitive sequence for detection of abnormality in the peripheral zone. However, DWI and quantitative ADC values were more useful for differentiating benign from malignant prostate lesions as well as low-grade from high-grade prostate carcinoma. This differentiation is important as management of benign prostatic lesions, low-grade prostate cancers, and high-grade prostate cancers vary. Granulomatous prostatitis lesions were hypointense on T2WI similar to carcinoma, but had higher ADC values.

DWI may help in differentiating benign scar, old hemorrhage or prostatitis from malignancy by showing diffusion restriction with a low ADC value in malignant lesions. Neoplastic proliferation causes increased cellularity and diffusion restriction of interstitial water molecules evidenced as hyperintense signal on DWI and hypointensity on corresponding ADC maps. As in our study, it has been shown that lesions with low-grade prostate cancer have a relatively high ADC compared to lesions with high-grade prostate cancer, which have aggressively proliferating cells.^[2,10-12] DWI is the most important sequence in the recent version of PIRADS for detection of peripheral zone prostate cancer.^{[2,6,7,10-14]}

In our study, the mean ADC values for high-grade prostate cancer, low-grade prostate cancer, prostatitis and normal prostatic parenchyma were increasing in trend which is seen in other studies as well. However, there is a difference in the mean values obtained for each category in other studies. Many studies in the literature have shown different ADC values for prostate lesions. Kim et al.^[14] found the mean ADC of prostate cancer to be 0.963 × 10⁻³ mm²/sec compared to the normal tissue 1.572 × 10⁻³ mm²/sec (P < 0.001). In a study by Tamada et al.,^[10] the peripheral zone prostate cancer had significantly lower mean ADC 1.02 ± 0.25 × 10⁻³ mm²/sec compared to the normal peripheral zone 1.80 ± 0.27 × 10⁻³ mm²/sec (P < 0.0001). Hambrock et al.^[15] have shown the correlation of ADC value with grade of prostate cancer. The median ADC value in low-grade prostate cancer and high-grade prostate cancer were 1.30 ± 0.30 × 10^{− 3} mm²/sec, and 0.94 ± 0.30 × 10^{− 3} mm²/sec, respectively. In another study by Nagel et al.,^[16] the median ADC value of normal prostate tissue, low-grade prostate cancer, and high-grade prostate cancer were 1.22 ± 0.21 x 10^-3 mm²/sec, 0.88 ± 0.15 x 10^-3 mm²/sec, and 0.88 ± 0.13 x 10^-3 mm²/sec, respectively. Their study did not show any significant difference of ADC between low-grade prostate cancer and high-grade prostate cancer. However, the ADC of prostate cancer was significantly lower than that of normal prostatic parenchyma. In a similar study by Pickles et al. (study done on 3T scanner), peripheral zone prostate cancer had lower mean ADC value of 1.38 ± 0.32 × 10⁻³ mm²/sec than noncancerous normal peripheral zone (1.95 ± 0.50 × 10⁻³ mm²/sec, P < 0.001).^[13]

A significant correlation between ADC value and histopathological grade of prostate cancer was found in our study. The ADC value in our study is lower than other previous studies described above. This may be due to the following reasons. First, ADC value depends not only on higher/lower field strength scanner (1.5T vs 3T) but also on manufacturer (e.g., Siemens, Philips). Second, interobserver variability in measuring ADC from the lesion also leads to variation. Since the lesion itself is not hom*ogeneous, the site on which the ROI is drawn and the area of ROI for measuring the ADC may lead to different measurements. Third, difference in b values taken may result in differences in ADC value.

T1WI is used to rule out hemorrhage that may be seen in recent prostate biopsy. Thus, a gap of 6 weeks is advised between previous biopsy and imaging to prevent false positive cases.

DCE MRI is a fast T1-weighted sequence to assess tumor angiogenesis. Malignant lesions show early and high-amplitude enhancement due to permeability of newly formed fragile vessels followed by rapid washout compared to normal prostatic parenchyma.^[17] DCE MRI images can be assessed qualitatively by simple visual analysis, semi quantitatively by looking at the type of curve and quantitatively by calculating the transfer constant (Ktrans) and the rate constant (Kep). There is a conflicting evidence regarding the role of DCE-MRI for the diagnosis of prostate cancer, with some studies showing a correlation while others show no correlation.^[3,18,19] In recently updated and widely used PIRADS 2 classification, only qualitative DCE-MRI analysis is used. It too has a limited role in the assessment of category 3 lesions, which are upgraded to assessment category 4 if contemporaneous early enhancement is present.^[20,21] With respect to DCE-MRI, our study found a positive correlation with a type 3 curve which was predominantly seen in high-grade prostate cancer. We made assessment qualitatively as well as semi quantitatively. Type 3 curve had relatively lower sensitivity, but higher specificity for detection of prostate carcinoma.

MRS is based on the detection of various cellular metabolite concentrations in prostate tissue. In cancer, choline is elevated due to increased cell-membrane turnover and rapid phospholipid metabolism while citrate signals decrease compared to normal tissue. Previous studies have shown a direct correlation of the Cho/Cit ratio with the histopathological grade of prostate cancer. In these studies, it was demonstrated that lesions having higher Gleason score on histopathology had higher Cho+Cr/Cit or Cho/Cit ratio. Studies have also shown an overlap between the Cho+Cr/Cit ratio in prostate cancer with benign prostatic hyperplasia and the non-cancer regions.^[22] The mean Cho/Cit ratio for high-grade prostate cancer was higher than that of low-grade prostate cancer and prostatitis in our study. However, a large standard deviation indicates a significant overlap and wide variation of values noted in all the three groups.

Chronic granulomatous prostatitis, which appears as a hypointense focus on T2-weighted imaging (T2WI), shows diffusion restriction with mild reduction in ADC, with moderate or marked enhancement on DCE-MRI. Due to similar imaging characteristics as prostate cancer, this could be a reason for fall in specificity of DWI. The Cho/Cit ratio may help in distinguishing chronic granulomatous prostatitis from high-grade prostate cancer. Choline is not elevated in chronic granulomatous prostatitis that differentiates it from prostate cancer. Chronic granulomatous prostatitis had a Cho/Cit ratio <0.50 similar to other studies.^[23-26]

Currently MRS is not included in prognostication of prostate cancer, as there is a wide variation in previously published studies and absence of a common agreed cutoff value. Second, there is issue of resolution of closely placed metabolites on the peak resonant frequency scale. Since creatine and choline are closely spaced on the ppm scale (creatine at 3.0 ppm, choline at 3.2 ppm), it was difficult to separate the two using previous generation 1.5T scanners. So, the choline and creatine peaks were calculated together as one. Moreover, due to the variation in citrate content in various parts of the gland and in population, the cho/cit ratio may not be a true indicator of the choline content in the region of interest.^[27] Lastly, absence of adequate fat saturation and water suppression may confound the results of spectroscopic assessment of Mp-MRI limiting its applicability for prostate cancer prognostication.

Targeted MRI-TRUS fusion biopsy of the suspicious lesion combined with standard biopsy cores has a significantly higher detection rate when compared to nontargeted standard TRUS guided biopsy.^[28]

Using T2WI, DWI, and DCE-MRI together, as in PIRADS version 2, the sensitivity, specificity, and positive predictive value for detection of prostate cancer in our study is comparable to other previous studies. In a meta-analysis by Hamoen EHJ et al.^[29] PIRADS 2 had a pooled sensitivity and specificity of 78% and 79%, respectively. Another meta-analysis of 13 studies by Zhang et al.,^[30] has shown that “PIRADS 2 has a pooled sensitivity of 0.85 (0.78–0.91) and pooled specificity of 0.71 (0.60–0.80). The positive predictive values ranged from 0.54 to 0.97 and negative predictive values ranged from 0.26 to 0.92.

We found DWI as the most sensitive sequence for detection of prostate cancer. Type 3 curve in DCE-MRI adds to the diagnosis in inconclusive cases. As the treatment for high-grade and low-grade prostate cancer is totally different, low grade prostate cancer managed conservatively while radical prostatectomy done for high-grade prostate cancer, it is important to differentiate these two. ADC value has a positive correlation with histopathology Gleason score grade of prostate cancer.

There are a few limitations. First, small sample size is the limitation of our study. Second, most of the cases included were positive for prostate cancer. This could be due to the selection bias as all of the patients with raised PSA level has positive DRE findings.

CONCLUSION

According to PI-RADS v2.1, DWI/ADC is the most important to evaluate the peripheral zone, although T2WI is sensitive for abnormality in that zone. DWI lesions can be useful to differentiate benign prostatitis from prostate carcinoma as well as low-grade from high-grade prostate carcinoma. ADC value has a positive correlation with histopathological grade of prostate cancer. Since high-grade and low-grade tumors are managed differently, ADC can be a surrogate marker for grade of prostate cancer. Although histopathology is the gold standard, Mp-MRI is an useful adjunct to localize the abnormal lesion and facilitate fusion biopsy to increase diagnostic yield. Further study with large sample size is needed.

Ethical statement

This study was approved by our institutional ethics committee.

Informed consent

Informed consent was obtained from all the patients included in the study.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.

Acknowledgements

We thank Dr Sanjeev Kumar Bhoi for statistical analysis.

REFERENCES