Does volumetric measurement of ADC values achieve higher diagnostic performance in bladder cancer MRI?

Abstract

Background

Apparent diffusion coefficient (ADC) value is an important part of bladder cancer magnetic resonance imaging (MRI) assessment and can predict the aggressive and invasive potentials. There is growing interest in whole tumor volume measurements.

Purpose

To investigate if the volumetric ADC measurement method will significantly exceed the diagnostic performance of the selected region of interest (ROI) method in everyday practice.

Material and Methods

A prospective evaluation was carried out of 50 patients with bladder cancer by two radiologists. The mean and the minimum ADC values were measured using both methods. The inter-reader agreement was determined by the intraclass correlation coefficient. The ADC values were compared between different grades, states of muscle invasion, and lympho-vascular invasion (LVI); then, validity was evaluated using receiver operating characteristic (ROC) curves. Areas under the curve (AUC) were then compared for the level of statistical significance.

Results

The inter-observer agreement was excellent for the ADC values using both methods. The volumetric measurement provides higher mean and lower minimum ADC values with statistically significant differences (P <0.00001). The highest diagnostic accuracy for differentiating tumor grade and predicting muscle invasion was for the minimum ADC by a selected ROI. However, the differences between the achieved AUCs were of no statistical significance. None of the ADC values predicted LVI with statistical significance.

Conclusion

The selected ROI and volumetric measurement methods of mean and minimum ADC in bladder cancer yield different values, still having comparable diagnostic performance with accurate ROI sampling. The minimum ADC value by ROI is preferred in everyday clinical practice.

Keywords

Bladder cancer magnetic resonance imaging apparent diffusion coefficient volumetric measurement

Introduction

A definitive preoperative diagnosis of bladder cancer (BC) depends on cystoscopy and tissue biopsy; however, the use of magnetic resonance imaging (MRI) as a non-invasive, initially guiding diagnostic modality with no radiation exposure is supported by clinical evidence (1,2). A useful functional quantitative tool on MRI is the apparent diffusion coefficient (ADC) value of diffusion-weighted imaging (DWI) showing lower values with higher tissue cellularity (3). With no need for a contrast agent, it has been an efficient biomarker for aggressive and invasive characteristics. Tumor aggressiveness, expressed by high tumor grade and high proliferative biomarkers such as Ki 67, p53, and p21 on immunohistochemistry, has been associated with low ADC values (4,5). Moreover, ADC values also correlate with the invasiveness of BC, which is reflected by higher stage and muscle invasion as well as lympho-vascular invasion (LVI) (4).

It is agreed that within heterogenous necrotic lesions, readings should be obtained from solid non-hemorrhagic parts of the tumors avoiding calcifications and artefacts, cross-referencing with the other sequences to properly delineate the lesions and their margins (6). In addition, the drawn region of interest (ROI) should not include the stalk for tumors having a non-restricted fibrovascular stalk on DWI (7). Both the mean and minimum ADC values have been studied for their validity in different organs (8,9).

Advances in software tools allowed incorporating the whole tumor as a volume of interest (VOI) and thus performing volumetric measurements on ADC maps. Moreover, this can be executed in a semi-automated or automated manner. It has been suggested that because of the heterogeneity of lesions, ROI methods may be more prone to sampling error and deficient evaluation of the whole lesion; thus, volumetric measurements can provide a more comprehensive assessment (10). Inter-observer variability is another important factor to consider in the assessment of a measurement tool to be used in routine reporting (11).

Given the need for special software and the time consumed for the placement of VOI, the aim of the present study was to investigate whether volumetric measurements significantly exceed the selected ROI methods in diagnostic performance or inter-observer agreement in everyday clinical practice.

Material and Methods

Patients

Between 2021 and 2022, 70 patients with sonographically detected bladder masses presented to our department for an MRI examination. We prospectively studied these patients, excluding 20 of them (for poor image quality, benign or inconclusive histopathological data). Consequently, 50 patients were enrolled in the study after providing written informed consent.

Approval for this study was obtained from the Ethics Committee of the Faculty of Medicine at Alexandria University (serial number: 0201508). The reviewer was Professor Dr Maha Ghanem, Chairman of the Ethics Committee of the Faculty of Medicine Alexandria University.

MR scanning protocol

After adequate bladder distension, all patients performed the examination on a 3.0-T MRI scanner (Ingenia; Philips MR Systems, The Netherlands) applying a surface coil (16-channel torso phased-array) in the supine position. The parameters used for the DWI sequence in the axial plane were as follows: matrix = 128 × 128; field of view (FOV) = 24 × 24 cm; repetition time (TR)/ echo time (TE) = 6159/81 ms; slice thickness = 3 mm with no gap; and four b-values (0, 500, 1000, and 1500 s/mm²). T2-weighted (T2W) imaging in the three planes and pre- and post-contrast T1-weighted (T1W) imaging were also acquired as part of the diagnostic procedure.

Image analysis

The images were analyzed with a DICOM viewer (Osirix v.5.6 64-bit; Pixmeo Sarl, Switzerland). Two radiologists with more than 5 years of experience in urogenital MRI reporting analyzed all the images independently. They reported the mean and minimum ADC values for all patients with the two methods, with each method in a separate setting, being blinded to patients’ clinical data, pathology reports, and each other's results.

Volumetric mean and minimum ADC were obtained by freehand ROIs drawn along the border of the tumor in each slice showing it, to cover the entire solid tumor volume excluding the stalk, hemorrhage, necrosis, and calcification. The whole-volume mean and minimum ADC values were then automatically calculated by the DICOM viewer. To measure the mean ADC by the selected ROIs, the mean value of three circular ROIs, placed within the most representative solid tumor areas, was calculated. The minimum ADC by a selected ROI was obtained by placing a small circular ROI (with an area in the range of 2–5 mm²) at the visually determined prominently restricted areas.

In cases of multiple tumors, one representative lesion was studied, being that of the highest stage on an imaging basis or the largest among lesions with similar radiological stages.

Histopathologic analysis

All patients had cystoscopic specimens. Three patients repeated it due to lack of muscle within the initial specimen and 13 patients underwent subsequent radical cystectomy. The tumors were described according to the World Health Organization 2016 classification as low- or high-grade along with the state of detrusor muscle invasion in all specimens. LVI was specified whenever present.

Statistical analysis

The statistical analysis was performed using SPSS software version 20 (IBM Corp., Armonk, NY, USA). Inter-observer variability for ADC measurements with both methods was assessed by analyzing the intraclass correlation coefficient (ICC; <0.50 = poor agreement; 0.50–0.75 = moderate agreement; 0.75–0.90 = good agreement; >0.90 = excellent agreement). Then, minimum and mean ADC values were averaged between the two readers for further analysis.

Comparison of the averaged values of mean and minimum ADC between both methods was carried out using Wilcoxon signed ranks test and a comparison of the four values between different grades, states of muscle invasion, and LVI was conducted using the Mann–Whitney U test. Receiver operating characteristic (ROC) curves were generated to evaluate the diagnostic performance of each value in determining the histologic grade, predicting the state of muscle invasion. The area under the curve (AUC) was considered to be the relative diagnostic accuracy. Then, we compared the AUCs in a pairwise fashion to determine if the differences were significant using MedCalc for Windows (MedCalc Software, Mariakerke, Belgium).

Results

There were 50 patients (46 men [92%], 4 women [8%]; mean age = 62.7 ± 11.4 years) included in the study. Multiple tumors were found in 26 (52%) patients. There were 34 (68%) high-grade tumors and 16 (32%) low-grade tumors; there were 26 (52%) muscle-invasive lesions and 24 (4%) non-muscle-invasive lesions. Examples are shown in Figs. 1 and 2. LVI was present in 11 (22%) specimens: six transurethral resection of bladder tumor (TURBT) specimens and five cystectomy specimens. There were 3 (27.3%) non–muscle-invasive bladder cancers (NMIBCs) and 8 (72.7%) muscle-invasive bladder cancers (MIBCs).

Fig. 1.

A low-grade NMIBC with central stalk. The mean ADC by 3 ROIs (0.835), minimum ADC by selected ROI (0.69), whole-volume mean ADC (0.89), and whole-volume minimum ADC (0.65). (a) Axial T2-weighted fast spin-echo imaging, (b) axial T1-weighted fat suppression sequence post-contrast, (c) axial DWI at b-value 1000, (d) axial ADC map, (e) tumor segmentation at one slice, (f) calculation of the volumetric values, (g) histopathological image showing low-grade papillary TCC (T), with no invasion of the detrusor muscle bundles (M) (40×), (h) a higher power view (200×). ADC, apparent diffusion coefficient; DWI, diffusion-weighted imaging; NMIBC, non–muscle-invasive bladder cancer; ROI, region of interest; TCC, transitional cell carcinoma.

Fig. 2.

A high-grade MIBC. The mean ADC by 3 ROIs (0.74), minimum ADC by selected ROI (0.45), whole-volume mean ADC (0.77), and whole-volume minimum ADC (0.13). (a) Axial T2-weighted fast spin-echo imaging, (b) axial T1-weighted fat suppression sequence post-contrast, (c) axial DWI at b-value 1000, (d) axial ADC map, (e) tumor segmentation at one slice, (f) calculation of the volumetric values, (g) histopathological image showing high-grade non-papillary TCC (T) invading the detrusor muscle bundles (M) (100×), (h) a higher power view of lympho-vascular invasion (arrows) (200×). ADC, apparent diffusion coefficient; DWI, diffusion-weighted imaging; MIBC, muscle-invasive bladder cancer; ROI, region of interest; TCC, transitional cell carcinoma.

The inter-observer agreement on the ADC values obtained from both measurement methods was excellent, as presented in Table 1. The best agreement was upon the whole-volume mean ADC value.

Table 1.

ICC for mean and minimum ADC by both methods.

	Mean ADC value		Minimum ADC value
	For 3 ROI values	For whole tumor volume	By a selected ROI	For whole tumor volume
Reader 1	0.800 ± 0.179	0.848 ± 0.181	0.583 ± 0.195	0.362 ± 0.289
Reader 2	0.803 ± 0.177	0.868 ± 0.188	0.574 ± 0.187	0.325 ± 0.285
ICC coefficient	0.944	0.974	0.956	0.963
95% CI	0.904–0.968	0.939–0.987	0.925–0.975	0.914–0.982
Level of agreement	Excellent	Excellent	Excellent	Excellent

P value is <0.001; P ≤0.05 is considered statistically significant.

ADC, apparent diffusion coefficient; CI, confidence interval; ICC, intraclass correlation coefficient; ROI, region of interest.

The mean values and standard deviation for each measurement are shown in Table 2 for the different tumor grades and states of muscle invasion. The volumetric method provides statistically significantly higher mean and lower minimum ADC values than the selected ROI method.

Table 2.

Values for mean and minimum ADC values (×10⁻³mm²/s) obtained by the selected ROI methods and the whole tumor volume method for different tumor grades and states of muscle invasion.

	HG (n = 34)	LG (n = 16)	MIBC (n = 26)	NMIBC (n = 24)
Mean ADC values
For 3 ROI values	0.725 ± 0.121	0.963 ± 0.165	0.723 ± 0.121	0.887 ± 0.187
For whole tumor volume	0.782 ± 0.137	1.020 ± 0.166	0.769 ± 0.137	0.955 ± 0.180
P value	<0.001	0.012	<0.001	0.001
Minimum ADC values
By a selected ROI	0.487 ± 0.101	0.774 ± 0.186	0.470 ± 0.100	0.696 ± 0.193
For whole tumor volume	0.204 ± 0.167	0.640 ± 0.256	0.189 ± 0.166	0.511 ± 0.294
P₀ value	<0.001	0.005	<0.001	<0.001

Values are given as mean ± SD. P value compares mean ADC using both methods in each grade and each state of muscle invasion. P₀ value compares minimum ADC using both methods in each grade and each state of muscle invasion. P ≤0.05 is considered statistically significant.

ADC, apparent diffusion coefficient; HG, high grade; LG, low grade; MIBC, muscle-invasive bladder cancer; NMIBC, non–muscle-invasive bladder cancer; ROI, region of interest.

The four calculated ADC values did not differ significantly between LVI-positive and LVI-negative cases, as demonstrated in Table 3.

Table 3.

Values for mean and minimum ADC values (×10⁻³mm²/s) obtained by the selected ROI methods and the whole tumor volume method for different states of LVI.

	LVI positive (n = 11)	LVI negative (n = 39)	U	P value
Mean ADC values
For 3 ROI values	0.799 ± 0.127	0.743 ± 0.199	205.5	0.84
For whole tumor volume	0.842 ± 0.134	0.810 ± 0.182	181	0.44
Minimum ADC values
By a selected ROI	0.541 ± 0.148	0.530 ± 0.196	195	0.66
For whole tumor volume	0.382 ± 0.258	0.273 ± 0.293	180.5	0.43

Values are given as mean ± SD. P value compares mean and minimum ADC using both methods between LVI-positive and LVI-negative cases. P ≤0.05 is considered statistically significant.

ADC, apparent diffusion coefficient; LVI, lympho-vascular invasion; ROI, region of interest; U, Mann–Whitney U test.

The mean and minimum ADC values obtained from both methods were higher in low-grade and NMIBC tumors compared to high-grade and MIBC tumors. ROC curve analyses to evaluate the diagnostic accuracy of each value in differentiating tumor grade and state of muscle invasion are shown in Table 4. In general, the AUC for the minimum ADC obtained by the selected ROI method was the highest to differentiate tumor grade (0.949) and to predict muscle invasion (0.882). However, pairwise comparisons of AUCs of the different values for the level of statistical significance revealed that there was no statistically significant difference between any value by any method and the others (p ≤ 0.05 was considered statistically significant).

Table 4.

Validity (AUC, sensitivity, specificity) for different parameters to differentiate between HG and LG cases and predict the state of muscle invasion.

Parameter	AUC	P value	95% CI	Cutoff*	Sensitivity (%)	Specificity (%)
Differentiating HG and LG
Mean ADC by 3 ROIs	0.888	<0.001^†	0.791–0.985	≤0.86	91.18	62.5
Mean ADC of whole tumor volume	0.881	<0.001^†	0.771–0.990	≤0.88	82.35	87.5
Minimum ADC by selected ROI	0.949	<0.001^†	0.893–1.006	≤0.60	94.12	87.5
Minimum ADC whole tumor volume	0.912	<0.001^†	0.825–0.998	≤0.38	85.29	81.25
Predicting the state of muscle invasion
Mean ADC by 3 ROIs	0.766	<0.001^†	0.632–0.900	≤0.76	73.08	79.17
Mean ADC of whole tumor volume	0.826	<0.001*	0.703–0.949	≤0.88	92.31	75.0
Minimum ADC by selected ROI	0.882	<0.001^†	0.783–0.982	≤0.55	84.62	83.33
Minimum ADC whole tumor volume	0.821	<0.001^†	0.707–0.935	≤0.31	80.77	70.83

*P ≤0.05 is considered statistically significant.

†

Cutoff was chosen according to Youden index (×10⁻³mm²/s).

ADC, apparent diffusion coefficient; AUC, area under the curve; CI, confidence interval; HG, high grade; LG, low grade; ROI, region of interest.

Discussion

The ADC value has been widely studied and correlated with the histopathological features of bladder cancer. Standard tumor features, such as the grade and stage (notably the state of muscle invasion), have been widely associated with the ADC value. Ki-67, as a proliferative marker for urothelial carcinoma, moderately correlated with the mean ADC in a large meta-analysis (12). Furthermore, the invasiveness pattern and lymphatic invasion showed significantly different median ADC values (13). Consequently, careful interpretation of this valuable imaging marker is needed to get the highest diagnostic benefit.

In our study, the mean ADC value by selected ROIs showed an AUC of 0.888 for differentiating low- and high-grade tumors; while for distinguishing MIBC from NMIBC, the AUC was 0.766. This was close to the results of the study by Rosenkrantz et al., who used the mean ADC value to predict the aggressiveness of urothelial carcinoma and showed an AUC of 0.902 to differentiate tumor grade and 0.765 to predict muscle invasion (14).

Concerning LVI, the ADC values in our study group did not differ significantly between LVI-positive and LVI-negative cases, which did not match the findings by Fujimura et al. (13). Although LVI also carries a prognostic utility and is included in the risk stratification of NMIBC, it has some limitations. These are the discrepancies between the results from TURBT and cystectomies (15) together with deficient reproducibility and reliability in the pathological assessment of LVI (16).

Previous studies showed that the minimum ADC value had better a diagnostic performance than other parameters, while other studies showed no significant difference (8,9). In our study, the minimum ADC value by a selected ROI achieved the highest AUC (0.949) to discriminate tumor grade as well as to predict muscle invasion, where the AUC was 0.882.

The heterogeneity of the bladder cancer lesions together with the trend of studying tumor tissue texture introduced the volumetric measurement. This has the benefit of avoiding sampling errors and the incomplete evaluation of different tumor regions (10). The whole-volume ADC metrics were studied in different malignancies, including the pancreas (17), rectum (11), cervix (18), ovaries (19), and prostate (20).

In accordance, bladder cancer whole tumor volume ADC metrics were studied by Rosenkrantz and colleagues to assess the aggressiveness of MIBC, concluding that the whole-volume mean ADC value showed a better association with tumor stage (AUC = 0.754) compared to other measurements (10). Recently, another study evaluated three methods of ADC measurement to reflect the grade and recurrence of bladder cancer (21). It showed that there was no significant difference in the mean ADC value measured by three ROIs, single section ROI and whole tumor volume methods (21).

Our study showed different values of mean and minimum ADC when obtained by the selected ROI method compared to the volumetric method; however, there was no statistically significant difference between their diagnostic performance (reflected by the AUCs) when compared together.

We also assessed the inter-observer agreement for both methods by calculating the ICC for the readings obtained by two radiologists specialized in genitourinary imaging. The ICC was excellent for both methods, with the highest ICC (0.974) obtained for the whole-volume mean ADC value, while the lowest (0.944) was for the mean ADC value by the selected three ROIs method, which is still excellent. Li et al., also proved excellent inter-observer agreement for the whole tumor volume mean ADC with ICC (0.90) but only good agreement (0.72) for the three ROIs method (21). Discrepancies in the experience of the observers can be the cause of such differences.

Although the whole tumor volume segmentation on the ADC map may provide more data by the histogram analysis, these are still under investigation for their diagnostic value. Concerns about the time consumed in the allocation of a VOI to accurately include the whole tumor volume are also present. This is notable for the manual placement and can be reduced by semi-automated tools; however, they are not widely available in routine clinical practice (22). This may drive radiologists to prefer the selected ROI methods in everyday reporting as long as the diagnostic performance is not affected.

The present study has some limitations. These include the relatively small sample size. The time consumed for the manual segmentation of the tumor to obtain the volumetric ADC values was long. It could have been reduced by using software tools for automated or semi-automated volumetric analysis; however, we preferred the manual method to attain the highest precision.

In conclusion, the volumetric and selected ROI methods yield different mean and minimum ADC values. However, there was no statistically significant difference in their diagnostic performance as well as their inter-observer agreement as long as the rules of correct sampling are followed. The minimum ADC value using the ROI method is preferrable in everyday use.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Nesma Elsayed Elshewy

Omnia Ezz Eldin

References

Messina

Pecoraro

Pisciotti

, et al. Seeing is believing: state of the art imaging of bladder cancer. Semin Radiat Oncol 2023;33:12–20.

Delli Pizzi

Mastrodicasa

Taraschi

, et al. Conspicuity and muscle-invasiveness assessment for bladder cancer using VI-RADS: a multi-reader, contrast-free MRI study to determine optimal b-values for diffusion-weighted imaging. Abdom Radiol 2022;47:1862–1872.

Sevcenco

Maj-Hes

Hruby

, et al. Apparent diffusion coefficient values obtained by unenhanced MRI predicts disease-specific survival in bladder cancer. Clin Radiol 2018;73:881–885.

Kobayashi

Takemura

Koga

. Apparent diffusion coefficient value as a biomarker for detecting muscle-invasive and high-grade bladder cancer: a systematic review. Appl Sci 2022;12:1278.

Sevcenco

Haitel

Ponhold

, et al. Quantitative apparent diffusion coefficient measurements obtained by 3-tesla MRI are correlated with biomarkers of bladder cancer proliferative activity. PLoS One 2014;9:e106866.

Yoshida

Koga

Kobayashi

, et al. Diffusion-weighted magnetic resonance imaging in management of bladder cancer, particularly with multimodal bladder-sparing strategy. World J Radiol 2014;6:344.

Kobayashi

Koga

Kajino

, et al. Apparent diffusion coefficient value reflects invasive and proliferative potential of bladder cancer. J Magn Reson Imaging 2014;39:172–178.

Sufana Iancu

Colin

Puech

, et al. Significance of ADC value for detection and characterization of urothelial carcinoma of upper urinary tract using diffusion-weighted MRI. World J Urol 2013;31:13–19.

Hirano

Satake

Ishigaki

, et al. Diffusion-weighted imaging of breast masses: comparison of diagnostic performance using various apparent diffusion coefficient parameters. AJR Am J Roentgenol 2012;198:717–722.

10.

Rosenkrantz

Obele

Rusinek

, et al. Whole-lesion diffusion metrics for assessment of bladder cancer aggressiveness. Abdom Imaging 2015;40:327–332.

11.

Lambregts

Beets

Maas

, et al. Tumour ADC measurements in rectal cancer: effect of ROI methods on ADC values and interobserver variability. Eur Radiol 2011;21:2567–2574.

12.

Surov

Meyer

Wienke

. Associations between apparent diffusion coefficient (ADC) and KI 67 in different tumors: a meta-analysis. Part 1: ADCmean. Oncotarget 2017;8:75434.

13.

Fujimura

Sakamoto

Sekita

, et al. Apparent diffusion coefficient value for estimating clinicohistological factors in bladder cancer including infiltration style and lymphatic invasion. Springerplus 2016;5:1–6.

14.

Rosenkrantz

Haghighi

Horn

, et al. Utility of quantitative MRI metrics for assessment of stage and grade of urothelial carcinoma of the bladder: preliminary results. AJR Am J Roentgenol 2013;201:1254–1259.

15.

Yoneda

Kamiya

Utsumi

, et al. Impact of lymphovascular invasion on prognosis in the patients with bladder cancer—comparison of transurethral resection and radical cystectomy. Diagnostics 2021;11:244.

16.

Mathieu

Lucca

Rouprêt

, et al. The prognostic role of lymphovascular invasion in urothelial carcinoma of the bladder. Nat Rev Urol 2016;13:471–479.

17.

Zhao

Ouyang

, et al. Quantified ADC histogram analysis: a new method for differentiating mass-forming focal pancreatitis from pancreatic cancer. Acta Radiol 2014;55:785–792.

18.

Downey

Riches

Morgan

, et al. Relationship between imaging biomarkers of stage I cervical cancer and poor-prognosis histologic features: quantitative histogram analysis of diffusion-weighted MR images. AJR Am J Roentgenol 2013;200:314–320.

19.

Kierans

Bennett

Mussi

, et al. Characterization of malignancy of adnexal lesions using ADC entropy: comparison with mean ADC and qualitative DWI assessment. J Magn Reson Imaging 2013;37:164–171.

20.

Donati

Mazaheri

Afaq

, et al. Prostate cancer aggressiveness: assessment with whole-lesion histogram analysis of the apparent diffusion coefficient. Radiology 2014;271:143–152.

21.

Liu

Ding

, et al. Quantitative assessment of bladder cancer reflects grade and recurrence: comparing of three methods of positioning region of interest for ADC measurements at diffusion-weighted MR imaging. Acad Radiol 2019;26:1148–1153.

22.

Ahlawat

Khandheria

Del Grande

, et al. Interobserver variability of selective region-of-interest measurement protocols for quantitative diffusion weighted imaging in soft tissue masses: comparison with whole tumor volume measurements. J Magn Reson Imaging 2016;43:446–454.