Measuring diagnostic accuracy of imaging parameters in pelvic lipomatosis

European Journal of Radiology 81 (2012) 3107–3114

Contents lists available at SciVerse ScienceDirect

European Journal of Radiology journal homepage: www.elsevier.com/locate/ejrad

Measuring diagnostic accuracy of imaging parameters in pelvic lipomatosis Yudong Zhang a , Shiliang Wu b , Zhijun Xi b,∗ , Xiaoying Wang a,∗ , Xuexiang Jiang a a b

Department of Radiology, Peking University First Hospital, Beijing, China Department of Urology, Peking University First Hospital, Beijing, China

a r t i c l e

i n f o

Article history: Received 15 December 2011 Received in revised form 23 May 2012 Accepted 25 May 2012 Keywords: Computed tomography (CT) Cystitis glandularis Hydronephrosis Magnetic resonance imaging (MRI) Pelvic lipomatosis

a b s t r a c t Objectives: To study whether the individual radiological findings can help predict diagnosis of pelvic lipomatosis (PL) or, specifically appreciate its progression. Methods: Data from 32 clinically proven cases of PL and 25 controls were collected. Two reviewers were recruited for a blinded evaluation, image features were recorded in terms of: (1) bladder shape; (2) bladder-rectosigmoid morphological indexes including ratio of superior–inferior to anterior–posterior length of bladder (SI/AP), angle between anterior and posterior wall (AAP), relative length of posterior urethra (rLPU), angle between bladder and seminal vesicle (ABS) and rectosigmoid morphological index (RMI); (3) secondary complications. Results were evaluated by an unpaired t test and ROC analysis. Results: The sensitivity and specificity were 40.6% and 100% for pear and banana-shaped bladder, 62.5% and 100% for SI/AP, 40.6% and 100% for AAP, 62.5% and 100% for ABS, 78.1% and 72% for rLPU, 59.4% and 96% for RMI, respectively. These radiological findings partially correlated with the severity of disease weighted by hydronephrosis and treatment grade. Image analysis demonstrated high prevalence of glandular cystitis (100%) and hydronephrosis (73.4%). Conclusion: We conclude that PL is a progressive disease involving multiple pelvic organs with high prevalence of intractable cystitis and hydronephrosis. The imaging characteristics can help predict diagnosis and, specifically appreciate progression. © 2012 Elsevier Ireland Ltd. All rights reserved.

1. Introduction Pelvic lipomatosis (PL) is a proliferative disease characteristic by overgrowth of normal fat in perivesical or perirectal areas [1]. Since it was initially described by Fogg and Smyth in 1968 [2], not more than 150 cases in the English literature had been reported. Despite its low incidence and benign features, more than 50% of patients are symptomatic because of mass sensation, compressive neuropathy or associated chronic cystitis [3,4]. Some patients will progressively develop obstructive hydronephrosis, and 40% of the patients at a mean period of 5 years after diagnosis of obstruction will progress into renal failure [5,6]. It is reported that more than 75% of patients with PL suffer as well the diseases of cystitis glandularis, cystitis cystica, or cystitis follicularis [7]. Moreover, it has been learned that proliferative diseases, especially adenomatous proliferation, are regarded as a potential precursor of adenocarcinoma [8].

An ability to make a specific diagnosis and more accurate evaluation of severity referring to renal failure and intractable cystitis/ureteritis without an invasive procedure would be quite helpful Moreover, even though the radiologic feature of PL has been discussed in the literature [9,10], the series are small, especially those that examine findings at CT and MRI [11]. Thus, the purpose of this study is to reevaluate the CT and MR features of PL, to determine whether there are radiologic characteristics that can help predict diagnosis of PL and, specifically, appreciate its progression. For this reason, (a) types of bladder shape, (b) a series of measurable characteristics of bladder and rectosigmoid, (c) secondary complications of urinary tracts and, (d) the correlation of individual radiological findings with severity of disease and management of the patients were evaluated respectively.

2. Material and methods 2.1. Patients

∗ Corresponding authors at: Department of Radiology, Department of Urology, Peking University First Hospital, No. 8, Xishiku Street, Xicheng District, Beijing 100034, China. Tel.: +86 01 66551122 2705; fax: +86 01 66551122 2705. E-mail addresses: pku [email protected] (Y. Zhang), [email protected] (S. Wu), [email protected] (Z. Xi), [email protected] (X. Wang), [email protected] (X. Jiang). 0720-048X/$ – see front matter © 2012 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ejrad.2012.05.031

This study was approved by local institutional review board. A computerized database was retrospectively searched from 2002 to 2011, for all cases of patients referred for CT and MR imaging evaluation of abdomen and cavitas pelvis in which the term “pelvic lipomatosis” or “glandular cystitis” appeared in the

3108

Y. Zhang et al. / European Journal of Radiology 81 (2012) 3107–3114

physician’s final assessment of the images. This search yielded a total of eighty-seven patients. A detailed medical record of these patients was also recalled from the medical record library. Patients with surgical procedure and who underwent at least six-month systematic clinical and radiographic follow-up after initial diagnosis of PL were included. This yielded records of a total of thirty-two cases. All were male, aged from 27 to 61 years with a mean age of 44 years. BMI of the patients ranged from 20.2 to 33.3 (median 24.5). More than 70% patients (23/32) had a five-month to five-year medical history of chronic abdominal pelvic pain or low urinary tract symptoms. Twenty-five additional cases (male; mean age, 44 years; range, 25–69 years) were retrospectively searched as a control. All of them underwent CT and/or MR imaging evaluation of abdomen and were without history of urinary diseases.

2.2. CT/MR study protocols In those thirty-two patients, nine underwent at least once CT of abdomen, twelve underwent non-contrast MRI and eleven underwent both CT and MRI. The abdominal CT examination was performed with a single-row (Plus 4; Siemens Medical Solutions, Forchheim, Germany), 16- or 64-detector row CT scanners (GE Medical Systems, Milwaukee, Wisconsin, USA). All patients were instructed to receive 500–750 mL of water orally 15–30 min before the examination. The protocol consisted of 3 phases of CT: un-enhanced, nephrographic, and excretory. For the singlerow CT, an un-enhanced CT scan from the kidneys to the urinary bladder was first obtained by using 120 kVp, and proximate 250 milliampere seconds (mAs), 1.0–1.5 pitch, 0.8 s/rotation, DFOV 35–40 cm2 , matrix 512 × 512, 5 mm section thickness. The kidneys were scanned again during the nephrographic phase at 80 s after the administration of 90 mL of 300 mg of iodine per milliliter intravenous contrast material (Ultravist 300; Bayer Healthcare Pharmaceuticals, Wayne, NJ; or Omnipaque 300, WinthropSterling, New York; Iopamidol 300; Bracco Diagnostics, Milan, Italy) at a speed of 2.5 mL/s with the same parameters. Then, patients received an intravenous drip infusion of 30–50 mL of normal (0.9%) saline. The abdomen/pelvis was rescanned during the excretory phase 5–10 min after the administration of contrast agent. The 16 or 64 detector row CT urographic protocols were similar to the single-row CT urographic protocol except for the use of 16/64 × 1.25 mm collimation, and minor variations in the tube current time product (100–250 mAs for un-enhanced CT, 100–300 mAs for nephrographic phase CT, and 110–310 mAs for excretory phase CT).

(a)

(b)

(c)

(d)

(e)

Fig. 1. Schematic diagram of bladder shape on median sagittal, coronal CT and MR images or 3D-CTU/MRU: (a) pear-shaped; (b) oval-shaped; (c) banana-shaped; (d) triangular-shaped; (e) irregular-shaped.

With regard to abdominal/pelvic MRI, the examination was performed on 1.0T (SMT-100X; Shimadzu Corp., Tokyo, Japan), 1.5T or 3.0T scanner (Signa ExciteTM ; GE Medical Systems, Milwaukee, Wisconsin, USA). The following sequences were collected on each subject during a single MR session: (a) respiratory-triggered fast spin-echo with T2-weighted imaging; (b) dual echo T1WI with breath hold; (c) single Shot Fast Spin (SS-FSE) with breath hold; (d) diffusion-weighted SE-echo echo-planar imaging with b factor of 800 s/mm2 ; (e) coronal 2D-SSFSE and/or 3D respiratory-triggered FSE with heavily T2 -weighted images, consisting of a maximum intensity projection (MIP) after imaging, to track the ureter and urinary bladder.

2.3. Image analysis Acquisition date and participant identification were removed from all images. The images were randomly allocated to one of two abdominal radiologists with seventeen and twenty years of experience in interpreting cross-sectional images. The reviewers were blinded to all clinical information and were asked to record the following imaging features: (1) bladder shape categorized as “pear,” “oval”, “banana”, “triangular” and “irregular” (Fig. 1); (2) bladder rectosigmoid morphological indexes (B-RMI) including the superior-inferior (SI), anteroposterior (AP) length of bladder and the ratio between of them (SI/AP), angle of anteroposterior bladder wall (AAP), relative length of posterior urethra (rLPU), angle of bladder-seminal vesicle (ABS) and rectosigmoid morphological index (RMI) (Fig. 2); (3) cystitis as evaluated by CT/MRI; (4) degree of hydronephrosis, according to the Society of Fetal Urology (SFU) Hydronephrosis Grading System.

Fig. 2. Demonstrative examples of measurements of B-RMI: (a) OS is the maximum length of bladder in median sagittal plane, CD and EF are the maximum distance from anterior and posterior wall to line OS, respectively, the ratio of superior–inferior to anteroposterior length of bladder (SI/AP) equals to OS/(CD+EF); angle between anterior and posterior wall of bladder (AAP) equals to ∠AOP; GH is horizontal line across the superior margin of pubic symphysis, OP is the distance from vesical neck to line GH, which is assigned to evaluate the relative length of posterior urethra (rLPU); (b) angle between bladder and seminal vesicle (ABS) in axial section (∠BOS); (c) AB is the horizontal line across basilar part of seminal vesicles in median sagittal plane and CD is the anteroposterior length of rectum, EF is the transverse diameter of rectum in axial-section, and the ratio of CD to EF is assigned as RMI.

Y. Zhang et al. / European Journal of Radiology 81 (2012) 3107–3114

3109

Fig. 3. (a) Normal fat distribution in pelvic cavity of a control patient as demonstrated on axial CT image; (b) T1 -weighted, axial MR image in one patient with PL shows excessive pelvic fat deposition, the bladder and rectum are compressed and deformed, abundant proliferated fibrous tissues (sign of “salt and pepper”) are detected in perivesical or mesorectal and the presacral areas (arrow); (c) vascular proliferation (arrow) can be seen around bladder on T2 WI with fat suppression.

2.4. Statistical analysis

3.2. Bladder shape

Statistical analysis was performed by using software packages (SPSS, SPSS, Chicago, III; Origin, Microcal, Northampton, Mass). The measurement data, including SI, AP, SI/AP, AAP, rLPU, ABS and RMI, was reported as mean ± standard deviation. An unpaired t test was performed in order to analyze the difference of SI, AP, SI/AAP, rLPU, ABS and RMI between PL and control group, P values of less than 0.05 were considered to indicate a statistically significant difference. Receiver operating characteristic (ROC) curves of SI, AP, SI/AAP, rLPU, ABS and RMI, featuring 1-specificity on the Xaxis and sensitivity on the Y-axis for the different thresholds, was calculated. The area under the ROC curves (AUCs), corresponding to the best thresholds of these imaging indicators, were determined. The sensitivity (SEN) and specificity (SPE) of each imaging parameter were calculated at the level of its best threshold. With regard to the enumeration data, such as the shapes of bladder, the radiological features of bladder wall and the degree of ureter obstruction, results were expressed as counts (or proportions in %). In order to determine whether individual radiological findings correlated to severity of disease, the patients were grouped into SFU 0–2 and SFU 3–4 based on hydronephrosis grade. Also two treatment-dependent groups were creased: one with a single aggressive transurethral resection of bladder tumor (TURBT) and another with TURBT plus double-J stent placement or with a surgical treatment as urinary diversion with ileum conduit. We hypothesize that a high SFU grade and management with TURBT plus D-J placement or urinary diversion other than with a single TURBT should be highly correlated to a high severity and poor prognosis of PL. Then a two-independent-sample t-test was employed to analyze the individual difference.

With regard to bladder shape, there was great difference between PL and controls (Table 1). It revealed that pear and bananashaped bladder were highly specific (100%), but relatively poorly sensitive (40.6%) to predict PL. The oval, triangular and irregularshaped bladders can be detected in either PL or controls (Fig. 4). 3.3. B-RMI Unpaired t test revealed a statistically significant difference of the imaging parameters, including SI, AP, SI/AP, AAP, rLPU, ABS and RMI, between patients with PL and controls. The B-RMI parameters with the maximal AUC, representing the most accurate imaging parameters to predict the patients suffering from PL, was AAP (AUC = 0.98) with an optimal threshold of 70◦ , the second most accurate parameter was RMI (AUC = 0.97) with a threshold of 1.25 and the third one was SI/AP (AUC = 0.96) with a threshold of 2. The parameters of ABS, SI, AP and rLPU also showed high AUC scores ranging from 0.74 to 0.94 (Fig. 5). The histograms of the best thresholds for each imaging parameters obtained by the ROC analysis was performed separately for each patient revealed a narrow distribution around the value of these thresholds obtained from the global analysis (Fig. 6). Taking the best thresholds of these imaging parameters as diagnostic criteria, all the parameters showed high specificity (72–100%) to predict standard diagnosis of PL. In view of sensitivity, the first susceptible parameter was rLPU (SEN = 78.13%), the second two ones were SI/AP and ABS (SEN = 62.50%) and the third one was RMI (SEN = 59.38%). But relatively low sensitivities were detected in parameters of SI, AP and AAP (Table 2). 3.4. Glandular cystitis

3. Results 3.1. Excessive fat deposit Generally in controls, the fat showed homogeneous low attenuation on CT images and hyper intensity on T1 -weighted MR images. In PL group, all the patients presented with excessive pelvic fat deposition, which was characterized with increased fat intensities on un-enhanced CT (< −100 HU) or T1 -weighted MR images, identified to the subcutaneous fat, in the perivesical and perirectal areas. 68.75% (22/32) of these patients suffered from various degrees of fibroplasia and proliferation, showing a characteristic sign of “salt and pepper” on T1 WI (Fig. 3).

Urodynamic examination of 26 patients (26/32, 81.25%) revealed an extreme obstructive pattern on uroflowmetry in combination with detrusor dysfunction. On CT and MR images, 90.63% (29/32) patients with PL were characterized with various degrees of Table 1 The changes in bladder shape of patients with PL. Bladder shape

Controls (n = 25)

PL (n = 32)

SEN

SPE

Pear Oval Banana Triangular Irregular

0 18 0 2 5

5 11 8 6 2

15.6% 34.4% 25% 18.8% 6.2%

100% 28% 100% 92% 80%

3110

Y. Zhang et al. / European Journal of Radiology 81 (2012) 3107–3114

Fig. 4. The changes in bladder shape of patients with PL: (a) intravenous urography (IVU) shows an inversed pear-shaped bladder combining with a moderate hydronephrosis; (b) un-enhanced CT with multi planar reconstruction (MRP) in median sagittal plane shows an inversed pear-shaped bladder, the elevated bladder neck and prolonged rectosigmoid; (c) the vertically oval-shaped bladder with thickened posterior wall of urinary bladder; (d) banana-shaped bladder with soft tissue mass intensity in trigonum vesicae, and prolonged posterior urethra; (e) the triangular-shaped bladder; (f) the irregular-shaped bladder.

the distal ureters were medially deviated, 7.81% (5/64) were laterally deviated, and the other 46.88% (30/64) were not markedly deviated. Twenty-two patients underwent a single TURBT, four underwent TURBT plus double-J ureteral stent placement and, six patients suffering from severe hydronephrosis and intractable cystitis/ureteritis received a surgical treatment as urinary diversion with ileum conduit, and following biopsy proved a fatty infiltration in lamina muscularis of involved ureters (Fig. 8). The correlation between radiological findings and clinical prognosis was demonstrated in Fig. 9. Results indicated that the mean value of SI/AP

thickened bladder wall and soft-tissue masses detected in trigone. 53.13% (17/32) patients presented with solitary or multiple diverticulums of the bladder (Fig. 7). 3.5. Hydronephrosis There was no hydronephrosis in controls. A high prevalence of hydronephrosis (47/64, 73.44) was revealed in total 64 ureters of 32 patients with PL (Table 3). The involved ureters were characterized with thickening of ureter wall, widely distributed soft tissue intensity in combination with narrowed lumen. 45.31% (29/64) of 1.0

1.0

rLPU RMI SI SI/AP ABS

0.6

Sensitivity

Sensitivity

0.8

0.4

0.2

0.8

0.6

0.4

0.2

0.0

0.0

0.0

(a)

AP AAP

0.2

0.4

0.6

1 - Specificity

0.8

1.0

0.0

(b)

0.2

0.4

0.6

0.8

1.0

1 - Specificity

Fig. 5. ROC analysis in the 58 subjects of PL and controls. The B-RMI parameters most accurately predicting the patients suffering from PL is AAP (AUC = 0.98). The second most accurate parameter is RMI (AUC = 0.97) and the third is SI/AP (AUC = 0.96). The parameters of ABS, SI, AP and rLPU also show high AUC scores ranging from 0.74 to 0.94.

Y. Zhang et al. / European Journal of Radiology 81 (2012) 3107–3114 8

0.0

PL(n=32)

12.5 10.0

Frequency

2.5

0

PL(n=32)

6

5.0 2.5 0.0 -20 -15 -10 -5 0

10

10

0 8

10

2

5

0

0 0

2

4

6

8 10 12 14 16 18 20

SI (cm) CG(n=25)

6

6

4

4

4

2

0 10

PL(n=32)

8

Frequency

6

8

4

2

2

2

0

0

0

140 130 120 110 100 90 80 70 60 50 40 30 20

6

4

(f)

left ABS (°)

12

0.0

14

AP (cm)

1.0

2.0

3.0

40

CG(n=25)

4.0

5.0

6.0

SI/AP

(d)

CG(n=25)

30 20 10

PL(n=32)

8

6

AAP (°)

10

0 10

4

(e)

8

2

6

0 20 40 60 80 100 120 140 160 180 200

6

10 8

PL(n=32)

4

(c)

10

PL(n=32)

0

2

8

0 10

0 20 15

6

8

2

PL(n=32)

CG(n=25)

5

4

(b) CG(n=25)

4

2 5 10 15 20 25 30

rLPU (mm)

(a)

15

2

4

7.5

Frequency

2

Frequency

Frequency

5.0

Frequency

4

7.5

20

CG(n=25)

6

Frequency

CG(n=25)

6

10.0

Frequency

CG(n=25)

12.5

3111

0 40

PL(n=32)

30 20 10 0

0

20

40

60

80

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

100 120 140

Right ABS (°)

(g)

RMI

(h)

Fig. 6. Histograms of the best thresholds for B-RMI parameters. It reveals: (1) a difference of the parameter distribution between the control group and PL; (2) a narrow distribution (optimal threshold) around the value of 13 cm in parameter SI, 4.5 cm in AP, 2 in SI/AP, 70◦ in AAP, 75◦ in ABS, 2.5 mm in rLPU and 1.25 in RMI, respectively, obtained in PL group.

Fig. 7. (a) Compressed and deformed bladder, combining with a decreased volume shown on axial CT; (b) the soft-tissue mass observed on bladder wall in coronal MPR CT image; (c) multiple diverticula of the bladder accompanied with dilated ureter tracts demonstrated on MR Urography (MRU).

Table 2 The results of B-RMI between PL and control group (mean ± standard deviation). BRMI

CG (n = 25)

SI (cm) AP (cm) SI/AP AAP(◦ ) Left ABS(◦ ) Right ABS(◦ ) rLPU (mm) RMI

8.89 8.67 1.22 143.71 49.86 47.74 1.26 1.04

± ± ± ± ± ± ± ±

3.20 1.85 1.11 23.81 13.16 14.58 3.83 0.32

PL (n = 32) 11.22 4.94 2.53 71.96 86.12 85.00 8.06 1.64

± ± ± ± ± ± ± ±

2.64 1.62* 1.09** 23.02** 22.82** 27.89** 7.06** 0.66*

AUC

Best Threshold

SEN

SPE

0.74 0.94 0.96 0.98

≥13 ≤4.5 ≥2 ≤70

25.00% 37.50% 62.50% 40.63%

92.00% 100.00% 100.00% 100.00%

0.90

≥75

62.50%

100.00%

0.80 0.97

≥2.5 ≥1.25

78.13% 59.38%

72.00% 96.00%

SEN, sensitivity; SPE, specificity. * Unpaired t test between PL and controls, P < 0.05. ** P < 0.01. Table 3 Hydroureterosis in patients with PL. Kidney

Grade 0

Grade I

Grade II

Grade III

Grade IV

Right Left Total

10 (15.63%) 7 (10.94%) 17 (26.56%)

3 (4.69%) 4 (6.25%) 7 (10.94%)

8 (12.50%) 5 (7.81%) 13 (20.31%)

3 (4.69%) 9 (12.50%) 12 (18.75%)

8 (12.50%) 7 (10.94%) 15 (23.44%)

3112

Y. Zhang et al. / European Journal of Radiology 81 (2012) 3107–3114

Fig. 8. (a) The MRU reveals a pear-shaped bladder, grade II hydronephrosis of the right kidney and grade 3 hydronephrosis of the left kidney, as well as a stenosis of bilateral distal ureters that are medially deviated (arrow); (b) another case shows pear-shaped bladder and grade 3 hydronephrosis of the right kidney and grade 3 hydronephrosis of the left kidney, the left distal ureter is laterally deviated (arrow); (c) curved-reformation shows thickened distal ureter, although mild dilation of ureter, remarkable soft tissue intensity can be seen in distal ureter and perivesical areas, implying a chronic proliferous inflammation (arrow); (d) biopsy reveals abundant fatty infiltration (arrow) in lamina muscularis of distal ureter in one patient receiving a surgical treatment as urinary diversion with ileum conduit.

measured in SFU 3–4 was statistically higher than that in SFU 0–2 (P < .05), the mean value of AAP in SFU 3–4 was lower than that in SFU 0–2 (P < .05). Although the results of ABS, rLPU and RMI did not reach a statistical significance (P = NS vs. SFU 0–2), a relatively higher value was found in SFU 3–4. The mean values of ABS and RMI in group with TURBT plus D-J placement or urinary diversion were statistically higher than those measured in group with single TURBT (P < .05), the mean value of AAP in group with TURBT plus D-J placement or urinary diversion was lower than that in group with single TURBT (P < .05). The results of SI/AP and rLPU did not reach statistical significance (P = NS vs. single TURBT), but a relatively higher value was found in TURBT plus D-J placement or urinary diversion. Thus, it was proven that an increased SI/AP, ABS, rLPU and RMI, as well as a decreased AAP, were positively correlated to a high severity and poor clinical prognosis of PL.

5.0

Progression

In this study, the imaging features of PL in 32 clinical proved cases were systematically reviewed. Up to now, this is a report with the biggest sample size. And with the usage of a ROC analysis, we illustrated that the individual radiological findings, in terms of types of bladder shape, the changes in B-RMI and the secondary compilations in urinary tract, can help predict diagnosis and, specifically, appreciate the progression of disease. PL was previously reported to be of relatively low incidence [12,13]. However, as often found incidentally in elderly men during evaluation of unrelated symptoms, the true disease prevalence of PL is probably underestimated. It has supported that there are two separate clinical entities, one affecting elderly men and seldom causing significant ureteral obstruction, the other affecting

P=.04

Low High

P=.02

4. Discussion

100.0

P=.01

Progression

Progression Low High

120.0

P=NS P=.05

Mean AAP (°)

Mean SI/AP

P=NS

3.0

2.0

0.0

60.0

Treatment

P=NS

12.5

Progression

2.5

100.0

80.0

60.0

20.0 Hydronephrosis

(b)

Low High

40.0

20.0 Hydronephrosis P=NS

(c)

Treatment

P=NS

Low High

Hydronephrosis

Treatment

Progression P=.01

Low High

2.0 10.0

Mean RMI

Mean rLPU (mm)

80.0

40.0

1.0

(a)

Mean ABS (°)

4.0

7.5

1.5

1.0

5.0 0.5 2.5 0.0

(d)

Hydronephrosis

Treatment

(e)

Hydronephrosis

Treatment

Fig. 9. The correlation between radiological findings and severity of PL. Black and white bar represents prognosis with low and high grade, which was weighted by SFU grade and strategies of treatment, respectively. It was predicted that an increased SI/AP, ABS, rLPU and RMI, as well as a decreased AAP, were positively correlated to a high severity and poor clinical prognosis of PL.

Y. Zhang et al. / European Journal of Radiology 81 (2012) 3107–3114

young men and more often producing hydroureteronephrosis and renal failure [5,14]. However, our study revealed the prevalence more commonly in young population. Moreover, isolated cases of associated young children and women have been reported [15,16]. Therefore, it prompts that the diagnosis of PL should not be ignored solely on the basis that the patient is too young. Certain studies suggest that the etiology of PL is possibly related to obesity [17], but only two cases in our study presented with a definitive obesity, and the BMI of our patients ranged from 20.2 to 33.3 (median 24.5) that did not largely exceed the normal standard of native group. Therefore, we conclude that the obesity is not the single factor responsible for the pathogenesy of PL. Moreover, we revealed unique features of PL in term of fat distribution and attenuation/signal intensity on CT/MR images. Unlike patients suffering from obesity whose excessive fat can be randomly distributed in belly, pelvic cavity or subcutaneous space, the excessive fat in patients with PL were primarily distributed in perivesical and perirectal areas, then, gradually spread upward to the abdominal cavity. In addition, pericystitis and periureteritis, characterized with fibroplasia and vascular proliferation, can be detected in more than 60% patients, which infer that the development of PL is not only associated with a quantitative increase in fatty tissue, but also highly related to a chronic proliferative inflammation in cavitas pelvis. It is generally accepted that bladder shape is a valuable characteristic indicative of PL [18,19]. Similar imaging signs were detected in our study. It prompts that the “pear” and “banana” bladder are specific in predicting those who became PL, but the sensitivity is relatively low (40.6%). The “oval-shaped” bladder found in 34.4% of PL group, could also be seen in most controls. But patients in PL mostly showed vertically oval-shaped bladder, while control cases mostly showed laterally oval-shaped bladder. The triangular or irregularshaped bladder is not specific to predict PL, which can be seen in either PL or controls. A series of measurable characteristics were followed in this study, and the SEN and SPE were determined. It shows that the rLPU presents with a high sensitivity (78.13%) but relative low specificity (72%) to predict PL, we suppose it is because the elevation of neck of bladder could be not only caused by excessive fat, but also the enlarged prostate in the patients who suffer from prostatic hyperplasia. The parameters of AAP, SI/AP and ABS are highly specific, but poorly sensitive to the change of bladder. RMI is highly specific (96%) but relatively poorly sensitive (59.38%) to the change of rectosigmoid. We suppose that the relatively low sensitivity of AAP, SI/AP, ABS and RMI possibly be associated with the progression of PL. As in the early stage of PL, although a pathological process has happened in the pelvic fatty tissue, the retentive resistance of bladder or rectosigmoid is stable enough to bear the compression from excessive fat that the bladder could presents with a relatively normal shape and the measurements of BRMIs are in a normal range. In this condition, the measurement of BRMI is inefficacious, but an empirical assessment of whether an excessive pelvic fatty tissue occurs may be helpful. Beside this, a six-month or more systematic clinical and radiographic follow-up is greatly necessary. Moreover, we propose that the optimal method to increase sensitivity to predict PL, especially in the patients with early stage, is based on a quantitative measurement of volume of pelvic fat. This requires a three dimensional imaging data and a precisely designed mathematical model. Limited by a retrospective study design here, we have not implemented the quantitative measurement of volume of pelvic fatty tissue. But this work can be followed with interest in the further study. In this study, all patients suffered from a various degree of proliferative cystitis. CT or MR images were characterized with a local or wide thickness in mucous membrane, solitary or multi bladder diverticulums and soft-tissue masses commonly seen in trigone of

3113

the bladder. Although this etiology has not been fully understood, several possibilities have been proposed, such as obstruction of the posterior urethra, chronic urinary tract infection because of the alteration of inner environment in bladder wall, which is favorable for the growth of proliferative tissue [20,21]. It is important that those proliferative diseases, especially in the patients with adenomatous proliferation, have been regarded as a potential precursor of adenocarcinoma. Although the incidence of adenocarcinoma is low and the progression of these proliferative lesions to adenocarcinoma are slow, the association between cystitis glandularis and adenocarcinoma should be paid more attention, and the patient with cystitis glandularis and pelvic lipomatosis should be followed cautiously to detect the associated adenocarcinoma and repeat transurethral resection and/or cup bladder biopsies during early follow-up period. Our results showed an extremely high incidence of hydroureter and ureteritis. This leads to 68.8% (22/32) patients undergoing the single TURBT, 12.5% (4/32) undergoing TURBT plus double-J ureteral stent placement and, 18.8% patients receiving the surgical treatment as urinary diversion with ileum conduit due to severe hydronephrosis, intractable cystitis or ureteritis. Also it showed that the radiological findings were positively correlated to the severity and clinical prognosis of PL that was weighted by SFU grade and strategy of treatment. Although some measurements did not reach a statistical significance, generally, the values of SI/AP, ABS, rLPU and RMI in high grade were higher and, AAP was lower than those in low grade. The pathogenesis of intractable hydronephrosis and ureteritis is not yet fully understood, however, at least two positive pathological processes have been revealed in our study: (1) the known mechanical pressure from excessive fat; (2) the proliferative inflammation combining with adipose infiltration in distal ureter tract, which suggest that hydroureterosis is a chronic, complex and progressive complication. 5. Conclusions The databases of PL in 32 clinical proved cases were systematically reviewed and some useful radiologic characteristics were illustrated. We conclude that PL is a chronic, complex and progressive disease involving multiple pelvic organs, and imaging characteristics, including the types of bladder shape, the changes of B-RMI and secondary complications in tubal bladder, can help predict diagnosis and, specifically appreciate the progression of PL. References [1] Kupelian S, Robinson MRG. Pelvic lipomatosis presenting as acute ureteric obstruction. British Journal of Urology 1976;48(5):389–90. [2] Fogg LB, Smyth JW. Pelvic lipomatosis – a condition simulating pelvic neoplasm. Radiology 1968;90(3):558–64. [3] Barry JM, Bilbao MK, Hodges CV. Pelvic lipomatosis – rare cause of suprapubic mass. Journal of Urology 1973;109(4):592–4. [4] Costa T, Fitch N, Azouz EM. Proteus syndrome – report of 2 cases with pelvic lipomatosis. Pediatrics 1985;76(6):984–9. [5] Radinsky SH, Vijungco JG. Bilateral ureteral obstruction caused by pelvic lipomatosis. Journal of the American Osteopathic Association 1978;77(11):865–7. [6] Crane DB, Smith MJV. Pelvic lipomatosis – 5-year followup. Journal of Urology 1977;118(4):547–50. [7] Kume H, Kume Y, Takamoto K. Achondroplasia associated with pelvic lipomatosis. Lancet 1999;353(9157):1017–20. [8] Heyns CF, Dekock MLS, Kirsten PH, Vanvelden DJJ. Pelvic lipomatosis associated with cystitis-glandularis and adenocarcinoma of the bladder. Journal of Urology 1991;145(2):364–6. [9] Hietala SO, Ghahremani GG, Faunce HF, Yaghmai I. Radiologic manifestations of pelvic lipomatosis. Radiology 1977;17(3):130–5. [10] Moss AA, Clark RE, Pepper HW, Goldberg HI. Pelvic lipomatosis – roentgenographic diagnosis. American Journal of Roentgenology 1972;115(2):411–9. [11] Demas BE, Avallone A, Hricak H. Pelvic lipomatosis: diagnosis and characterization by magnetic resonance imaging. Urologic Radiology 1988;10(4):198–202. [12] Old WL, Stokes TL. Pelvic lipomatosis. Surgery 1978;83(2):173–80. [13] Miglani U, Sinha T, Gupta SK, et al. Rare etiology of obstructive uropathy: pelvic lipomatosis. Urologia Internationalis 2010;84(2):239–41.

3114

Y. Zhang et al. / European Journal of Radiology 81 (2012) 3107–3114

[14] Carpente Aa. Pelvic lipomatosis – successful surgical treatment. Journal of Urology 1973;110(4):397–9. [15] Joshi KK, Wise HA. Pelvic lipomatosis – 9-year follow-up in a woman. Journal of Urology 1983;129(6):1233–4. [16] Kumar A, Zaman W, Singh V, Kumar B, Srivastava A, Waklu AK. Pelvic lipomatosis in a child. Urologia Internationalis 2002;69(3):238–40. [17] Morettin LB, Wilson M. Pelvic lipomatosis. American Journal of Roentgenology, Radium Therapy, and Nuclear Medicine 1971;113(1):181–4. [18] Becker JA, Weiss RM, Schiff M, Lytton B. Pelvic lipomatosis. A consideration in diagnosis of intrapelvic neoplasms. Archives of Surgery-Chicago 1970;100(1):94–6.

[19] Martelli A, Bercovich E, Soli M, Daniele C. Pelvic lipomatosis – non-surgical diagnosis. European Urology 1979;5(6):369–70. [20] Andrianne R, Leruth E, Coppens L, Bonnet P, Waltregny D, de Leval J. Pelvic lipomatosis associated with glandular cystitis. Report of two cases. Progrès en Urologie 2005;15(1):81–4. [21] Masumori N, Tsukamoto T. Pelvic lipomatosis associated with proliferative cystitis: case report and review of the Japanese literature. International Journal of Urology 1999;6(1):44–9.