TSI ultrasound elastography for the diagnosis of chronic allograft nephropathy in kidney transplanted patients.

PURPOSE: To answer the question whether the TSI (tissue strain imaging) sonoelastography technique can contribute to the diagnosis of chronic renal allograft damage.
MATERIAL AND METHODS: A prospective study of 112 patients between June 2010 and April 2011 was conducted to compare elastography data with biopsy results and laboratory parameters in order to determine whether any correlations exist. Elastography parameters were acquired with a high-end ultrasound system and analyzed using the semiquantitative strain ratio. For comparison, patients were divided into three groups based on biopsy findings (Banff classification): group A: biopsy not necessary; group B: Banff grade I; group C: Banff grades II and III. Correlations were assessed by means of correlation (Pearson) and regression analysis. Differences between ordinal groups were tested for statistical significance by the Mann-Whitney U test.
RESULTS: Mean patient age was 54.2 ± 15.01 years. Fifty-nine percent of the patients were male. The calculated TSI strain ratio of groups A and C differed significantly (p = 0.024). Groups B and C (p = 0.056) and groups A and B (p = 0.88) showed no significant difference. The TSI strain ratio did not correlate with glomerular filtration rate (r = 0.105) or creatinine (r = 0.092).
CONCLUSION: The TSI sonoelastography technique can contribute to the differentiation of different stages of renal graft damage (according to Banff classification). However, significant results were not observed for all investigated features. The TSI technique should be further evaluated in future studies including larger numbers of patients.

Introduction

In recent years, there has been an increasing scientific interest in sonoelastography as a noninvasive diagnostic tool for detecting tissue abnormalities. Measurement of elastic properties in different body regions can provide information on pathologies ranging from inflammatory processes to malignant tumors(. Different technical approaches of sonoelastography have been pursued and shown promising results.

Examples include the diagnosis and characterization of lesions in the breast or prostate( and the diagnostic evaluation of liver pathologies(. Recently, a controversy has arisen as to whether sonoelastography is also suitable for identifying graft failure in kidney transplant recipients(.

So far, only sonographic techniques that are based on shear wave technology have been used to assess elastography data of the kidney including the sonographic technique of transient elastography (TE) and the technique of acoustic radiation force impulse (ARFI). Both techniques have been mainly used for assessing liver fibrosis.

Currently, a variety of diagnostic tools are in use to identify renal graft dysfunction following transplantation. Ultrasound with contrast medium has been found useful for evaluating graft function in transplant recipients(. The most widely used routine diagnostic techniques are laboratory tests and (Doppler) ultrasound(. The latter can reveal morphologic abnormalities and provides information on graft perfusion (resistance index, RI). In patients with progressive graft deterioration in serial examinations, biopsy continues to be the gold standard for confirming or ruling out suspected graft failure. A biopsy is invasive and carries some risk. It would therefore be very helpful to have a valid noninvasive diagnostic alternative for assessing renal transplant function. Sonoelastography might be a promising candidate.

The aim of the present study was to evaluate the potential of tissue strain imaging (TSI) – an offline sonoelastography technique – for detecting chronic renal allograft damage. We compared renal elastography findings with laboratory parameters and biopsy results in renal transplant recipients and tested for correlations.

Material and methods

Between June 2010 and April 2011, a total of 122 renal transplant recipients followed up by the interdisciplinary aftercare service (radiology/nephrology) of Charité – Universitätsmedizin Berlin were included in the study. Transplantations took place between 1991 and 2010. Hence, the follow-up included examinations of one-yearold to 19-year-old transplants. The patients who were included had to consent to ultrasound elastography and storage of data. Adequate TSI elastography was accomplished in 112 cases, meaning that the colored overlay display was complete within the region of interest (ROI) and the elastogram reflected the whole selected region within the parenchyma and calyces respectively (see fig. 1). Two patients did not match these prerequisites due to obesity (BMI > 35 kg/m2) and TSI elastography was not possible in these cases. In four patients, the system's integrated display of transducer pressure revealed that the pressure applied was not constant; these cases were excluded from analysis. In another four cases, patient files in the nephrology database were incomplete. As stated, a total of 112 cases could finally be included in the analysis.

Fig. 1 A

Display without colored overlay of elastography properties

TSI elastogram: ROI in renal pelvic-calyceal system (1) and renal parenchyma (2).

* Sinusoidal compression gradient. The yellow rectangle marks the moment of elastography measurement (during decompression).

# Displays consistent pressure application during compression time.

▲ Dimensionless elastography value in renal pelvic-calyceal system.

▼ Dimensionless elastography value in renal parenchyma.

Display with colored overlay of elastography properties

Ultrasound examination

TSI was performed using a high-end ultrasound system (Aplio XG SSA-790A, Toshiba, Otawara, Japan) with an 8 MHz PLT 805 AT linear transducer. B-mode image quality was optimized using integrated options for improving axial and lateral resolution and reducing noise (tissue harmonic imaging and standardized preset options). All examinations were performed by the same investigator (JK).

First, the transplant kidney was identified in the longitudinal B-mode view, which was followed by ten cycles of controlled compression and release. The sonographic raw data acquired during these cycles were stored on the system's hard disc for offline processing. The TSI technique does not allow real-time assessment of tissue elasticities. The examiner places regions of interest (ROI) in the target area, for which the system then generates an elastogram with color-coded elasticity information that is superimposed on the B-mode image. The ROIs are two large rectangles covering a representative parenchymal region (approx. 4 cm2 of size) and a second representative area of the calyceal system (approx. 2 cm2 of size). The size of all evaluated ROIs was similar and varied within a range of max. ± 10% (in each the parenchymal and the calyceal ROI). This analysis included only those cases in which the pressure display confirmed that uniform transducer pressure had been applied and the compressionrelease cycles resulted in a sinus-shaped curve (see fig. 1). Elasticity information was recorded during the release phases in order to standardize the point of time where elastography data was collected and to be able to better compare interindividual results. At the time of TSI, the examiner was unaware of biopsy findings and laboratory parameters. For a detailed description of the TSI technique, the reader is referred to an article that appeared in “Academic Radiology” in 2007(.

Data analysis

When interpreting an elastogram, one must be aware that the colors display relative elasticities of the target tissues rather than absolute values. This means that, in each examination, the whole spectrum of colors from red to blue may be present in the target anatomy(. For comparison of findings in different patients, ratios were calculated of calyceal elasticity to parenchymal elasticity. This so-called strain ratio is a measure of the difference in elasticity between these two tissues of the kidney and allows interindividual comparison of the relationship of calyceal elasticity to parenchymal elasticity:

The strain ratio is the reference measure of sonographically determined tissue elasticity and is used for comparison with biopsy findings and laboratory parameters(.

Histology

In 19 cases, biopsy findings were available for comparison. Biopsies were obtained with a high speed Bard MAGNUM biopsy system (Bard, Tempe, AZ, USA) with a 14 g × 16 cm biopsy needle.

Renal allograft fibrosis was graded using the Banff classification (see tab. 1)( of renal transplant pathology and was used for assigning the patients to one of three groups (group A: biopsy not necessary, n = 94; group B: Banff grade I, n = 16; group C: Banff grades II and III, n = 3) (see tab. 3). Patients in group A had an uneventful clinical and ultrasound-morphological course with stable allograft function, and no biopsy was required. The probability of fibrosis in this group is low.

Tab. 1

Banff classification(

Banff classification
Group I	Mild interstitial fibrosis and tubular atrophy (< 25% of cortical area), no evidence of any specific etiology (may include nonspecific vascular and glomerular sclerosis)
Group II	Moderate interstitial fibrosis and tubular atrophy (26–50% of cortical area), may include nonspecific vascular and glomerular sclerosis (as above)
Group III	Severe interstitial fibrosis and tubular atrophy/loss (> 50% of cortical area), may include nonspecific vascular and glomerular sclerosis (as above)

Tab. 3

Groups according to Banff classification and TSI elastography results

Group	Banff classification	n	Mean TSI strain ratio ± SD	TSI calyceal values ± SD	TSI parenchymal values ± SD
A	Patients without biopsy	94	2,34 ± 1,29	0,054 ± 0,028	0,023 ± 0,013
B	Patients Banff I	16	2,14 ± 0,78	0,045 ± 0,026	0,021 ± 0,008
C	Patients Banff II and III	3	1,14 ± 0,37	0,032 ± 0,014	0,029 ± 0,027

Statistical analysis

The data were analyzed using SPSS Statistics, version 19 (SPSS Inc., Chicago, IL). Associations between elastography parameters and biopsy results or functional laboratory parameters were evaluated using Pearson correlation and regression analysis. Descriptive statistical data are given as means and standard deviations and represented as box-and-whisker plots. The box contains the first and third quartiles (interquartile range, IQR). The line in the box indicates the median. The whiskers correspond to the 1.5-fold IQR. Values outside this range are marked. Differences in ordinal data were tested for significance using the Mann-Whitney U test. Multivariate analysis was performed using a linear model and taking covariates into account. Statistical significance was assumed at p values < 0.05.

Results

Elastography data from a total of 112 patients were included in the analysis. The patients had a mean age of 54.2 ± 15.01 years. Fifty-seven percent were male; 72.3% of the recipients had received kidneys from postmortem donors. The patient data are summarized in tab. 2.

Tab. 2

Overview of patient population

	Minimum	Maximum	Mean ± SD
Patient age (years)	19	79	54,23 ± 15,01
Patient sex			59% male
Donor type (deceased or living)			72.3% deceased donors
Transplant age (years)	9	77	50,17 ± 15,75
Time between transplantation and ultrasound examination (days)	72	7077	1510 ± 1690
Cold ischemia time (min)	30	1560	590 ± 363
Mixed ischemia time (min)	25	110	53,5 ± 15,8
GFR (ml/min)	4,98	138	45,26 ± 20,42
Creatinine (mg/dl)	0,64	9,13	1,90 ± 1,16
Transplant size (cm)	9	14	11,32 ± 1,25
Resistance Index (RI)	0,50	1,00	0,73 ± 0,09
TSI strain ratio	0,71	8,77	2,25 ± 1,22

Mean TSI strain ratios by Banff group are presented in tab. 3.

There was a significant difference in TSI strain ratios between groups A and C (p = 0.024). The difference remained significant after multivariate correction for the following factors: interval between transplant operation and time of ultrasound examination, pole distance, cold ischemia time, mixed ischemia time, mode of kidney donation, age of donor, and resistance index (RI). The differences between groups B and C (p = 0.056) and groups A and B (p = 0.88) were not significant (see fig. 2).

Fig. 2

Box-whisker-plot of TSI strain ratio regarding the different biopsy groups. There is a significant difference between groups A and C (p = 0.024). No significant difference was shown between groups B and C (p = 0.056) and groups A and B (p = 0.88)

There was no significant correlation between the TSI strain ratio and serum creatinine (r = 0.092 with p = 0.167) or between the TSI strain ratio and glomerular filtration rate (GFR) (r = 0.105 with p = 0.136). Descriptively, however, the point diagram indicated a trend: an increasing TSI strain ratio appeared to be associated with a decrease in creatinine levels and increase in GFR (see figs. 3 and 4).

Fig. 3

Scatter chart TSI strain ratio vs. serum creatinine

Fig. 4

Scatter chart TSI strain ratio vs. glomerular filtration rate (GFR)

Discussion

To our knowledge, this is the first study that examines renal allografts using TSI elastography. Several recent studies have shown the benefit of sonoelastography in other organs, particularly for the detection of cancer of the breast( and of the prostate(. Promising results with sonoelastography have also been obtained in grading liver fibrosis in patients with chronic hepatitis(. Whether sonoelastography is also useful for evaluating the degree of fibrosis in the assessment of renal graft function is still a controversial issue.

In a study of 20 biopsied patients conducted in 2009, Arndt et al.( showed a significant correlation between the degree of fibrosis of kidney grafts and sonographic elastography values. The authors also found significant differences in elastography values between different groups differentiated on the basis of GFR. They employed the sonographic technique of transient elastography (TE), which relies on shear waves and is different from the TSI technique used in our study. So far, TE has been mainly used for assessing liver fibrosis. Another difference between the two techniques is the choice of target region investigated. The technique of TE precludes the choice of a specific ROI by the examiner. The sampling box – meaning the area covered by the examination – is a predefined rectangle measuring approx. 10 × 25 mm; however, its exact position can only be roughly estimated.

Similar results were reported by Stock et al. for the acoustic radiation force impulse (ARFI) technique – another technique based on shear waves – in 2010 and 2011. In these studies of 18( and 8( patients, the authors found significant differences in elastography values determined by ARFI between different groups of transplant recipients classified according to biopsy findings.

Syversveen et al.( did not confirm these results. They found no significant differences in ARFI elastography results between different Banff groups.

A recent study in a rat model( found no clear correlation between the degree of renal parenchymal fibrosis and sonographic elasticity values.

We found marked differences only for some of the patient groups distinguished on the basis of biopsy findings. In general, the mean TSI strain ratio decreases as the degree of fibrosis increases (higher Banff groups).

It follows from this observation and the definition of strain ratio that, with increasing severity of fibrosis, either parenchymal elasticity increases or calyceal elasticity decreases. This observation might appear surprising as we would expect to observe a decrease in renal parenchymal stiffness as the severity of fibrosis increases. Actually, however, parenchymal elasticity was found to be relatively stable or even tended to increase slightly. Instead, we found a more marked mean decrease in calyceal elasticity with higher Banff groups. This explains the decreasing strain ratios, which is nearly one in biopsy group C (Banff II and III). Hence, with increasing fibrosis (on the basis of histology), there is an approximation of calyceal and parenchymal elasticity measured by sonoelastography. However, it remains open whether this finding reflects a true decrease in calyceal tissue. It is also conceivable that hardening of the parenchyma with increasing fibrosis may impair the sonoelastic evaluation of the underlying calyceal tissue. It is possible that shrinkage of the fibrotic parenchyma impairs pressure coupling to the calyceal system, resulting in underestimation of calyceal elasticity.

Banff classification not only shows the grade of fibrosis, but also takes inflammatory changes of the tissue into account. Corollary, a diffuse tissue hardening might be the result of an inflammatory process that can't be distinguished from fibrosis of other pathological reasons using sonoelastography.

Overall, the decreasing strain ratio suggests that a higher Banff group is associated with a change in the sonographic elasticity of the entire kidney graft. At present, it remains open whether the observed change is due to an inherent artifact of sonoelastography.

The differences in the pairwise group comparisons are not always statistically significant. This is due to the small numbers of cases, especially in group C (Banff II and III). A significant correlation with clinical parameters – serum creatinine and estimated GFR – cannot be established; however, there is a trend toward lower TSI strain ratios with higher serum creatinine levels and lower GFR.

Limitations

Acceptable biopsy results were available for only 19 of the 112 renal graft recipients included in the analysis of sonoelastography. This limits our results with regard to a possible correlation between biopsy results and sonoelastography data.

Sonographic elastography is not yet a standardized diagnostic procedure. Nevertheless, an attempt was made to perform the elastography examination in a reproducible manner, but deviations from the intended procedure cannot be fully excluded. One factor is that pressure with the transducer is applied manually and hence the amount of pressure is examiner-dependent. While this allows easy clinical handling of the technique, it precludes the application of controlled pressures as it would be possible in an experimental setting.

Conclusions

All results considered, our findings suggest that offline analysis of tissue elasticity measured by TSI elastography in kidney grafts can contribute to the grading of fibrosis compared with biopsy results. However, the small number of patients for whom biopsy results were available does not allow definitive correlation of sonoelasticity for all stages of fibrosis verified by biopsy. The results are promising and larger studies should be conducted to further explore the potential of this easy to handle and noninvasive technique in the clinical setting.