The Role of Host Factors and Bacterial Virulence Genes in the Development of Pyelonephritis Caused by Escherichia coli in Renal Transplant Recipients

Abstract

Background and objectives: The aim of this study was to determine the role of host factors and bacterial virulence genes in the development of pyelonephritis caused by Escherichia coli in renal transplant (Tx) recipients.

Design, setting, participants, & measurements: A total of 328 E. coli isolates from cases of cystitis (Cys; n = 239) or pyelonephritis (PN; n = 89), with 169 from renal Tx recipients, were subjected to molecular analyses to identify P-fimbria subunits (PapC, PapG II, and PapGIII), G- and M-fimbriae, and aerobactin. The presence of antibiotic resistance was also determined. Parameters such as gender, age, immunosuppression regimens, causes of ESRD, kidney donor, intraoperative anastomosis, use of double J stent, trimethoprim/sulfamethoxazole (TMP/SMZ) prophylaxis, and time after Tx were evaluated.

Results: A multivariate analysis showed a significant association between PN and renal Tx. In renal Tx recipients, the risk of occurrence of PN was significantly higher among males and for those no longer receiving TMP/SMZ prophylaxis. E. coli strains isolated from PN presented a lower prevalence of papGIII and lower rates of resistance to pipemidic acid. Although papGII was more prevalent in PN than in Cys, it was not independently associated with PN.

Conclusions: These findings suggested that renal Tx increases the risk for PN, and the male sex represented a host factor independently associated with risk, whereas the prophylaxis with TMP/SMZ was protective. The lack of papGIII and low resistance to first-generation quinolones were bacterial-independent risk factors for PN in Tx.

Urinary tract infections (UTIs) are one of the most common infectious diseases encountered in clinical practice. Especially in the setting of renal transplantation, asymptomatic bacteriuria or symptomatic forms such as cystitis (Cys) and pyelonephritis (PN) account for approximately 47% of infectious complications (1) and up to 60% of bacteremias (2). The development of UTIs depends on anatomical factors, the integrity of host defense mechanisms, and the virulence of the infecting organisms (3). Potential predisposing host factors in renal allograft recipients include immunosuppression, female sex, diabetes mellitus, urinary tract abnormalities, and urinary stents, among others (4).

Escherichia coli is the most common gram-negative microorganism identified in UTIs, even in transplanted patients (5,6). Uropathogenic E. coli adhere to uroepithelial cells through fimbrial adhesins, including type I, P-, M-, and G-fimbriae, and adherence is one of the most important steps in the development of UTIs (7). P-fimbria, composed of several protein subunits and encoded by pap (PN-associated pili) genes, is mostly associated with acute PN (8). PapG adhesins at the tip of P-fimbriae mediate specific binding to glycolipid receptors on the uroepithelia and renal tissue; three molecular variants of PapG have been identified (9,10). PapGII (11,12) and PapC (5,13) are suggested to be associated with upper UTI (PN), whereas PapGIII predominates in lower UTI (Cys) (14). In addition to adhesins, other virulence factors (i.e., aerobactin, hemolysin, and capsular polysaccharide) may be associated with the clinical manifestations of PN (12,13,15).

The acquisition of antibiotic resistance may be associated with phenotypic changes in bacteria that render them more virulent (16). On the other hand, other investigators have suggested that quinolone resistance involves a genotypic change associated with the loss of virulence factors in E. coli (17–20).

The aim of the study presented here was to evaluate the association of host and bacterial virulence factors with the development of PN by E. coli in renal transplant recipients.

Materials and Methods

Three-hundred ninety-six E. coli isolates from adult patients with UTIs referred to the Microbiology Laboratory of Federal University of São Paulo from February 2007 through November 2008 were prospectively submitted to molecular analyses.

Diagnostic Criteria and Patient Selection of E. coli UTIs

The selection of patients was based on a bacterial count ≥10⁵ CFU/ml of E. coli in the urine culture irrespective of the results of the urinalysis and a sufficiently detailed medical record available for a retrospective analysis. A clinical diagnosis of PN or Cys, a prerequisite for enrollment in the study presented here, was made in a total of 328 patients, from which 169 were renal transplant (Tx) recipients. Patients presenting with prostatitis (n = 5), asymptomatic bacteriuria (n = 19), neurogenic bladder (n = 6), and actinic Cys (n = 1) were excluded. Further recurrent episodes of UTI occurring in the same patients were also excluded from the analysis presented here (n = 37).

Written consent was obtained from all subjects, and the study was approved by the local ethics committee of the Federal University of São Paulo. Cys was clinically defined by the presence of dysuria, urinary urgency, and frequency. Acute PN was defined by the presence of fever (>38°C axillary) plus flank pain, with or without dysuria or urinary frequency. In renal Tx patients, PN was defined by the presence of fever and/or an increase of serum creatinine ≥20% over baseline values (21) on the occasion of the positive urine culture. Other accompanying symptoms or signs of PN included renal allograft tenderness, bacteremia, or a renal allograft biopsy specimen confirming bacterial PN. To avoid possible confounding factors, patients with increased serum creatinine due to acute cellular rejection, urinary tract obstruction, or immunosuppressant-associated nephrotoxicity at the time of their UTI were excluded. A decrease in serum creatinine or a negative urine culture, available soon after antibiotic therapy, further confirmed the presence of PN.

The parameters analyzed in the medical records were gender; age over 60 years old; previous episodes of UTI; specific causes of ESRD such as diabetes mellitus, nephrolithiasis, polycystic kidney disease (PKD), and vesicoureteral reflux (VUR) to native kidneys. Information about the type of kidney donor (living or deceased), the type of anastomosis (Gregoir or Politano–Leadbetter), double J stent use, the time posttransplantation, and the number of patients receiving trimethoprim/sulfamethoxazole (TMP/SMZ) prophylaxis and immunosuppression regimens were also sought. According to the routine of the Renal Transplant Unit, immunosuppression induction was performed in the cadaveric renal transplantation with basiliximab; thymoglobulin; or more rarely with daclizumab, muromonab-CD3, or alemtuzumab. Immunosuppression maintenance regimens consisted of (1) triple associations with calcineurin inhibitor (tacrolimus [Tac] or cyclosporine [CsA]), prednisone (Pred), and mycophenolate agents (mycophenolate mofetil [MMF], mycophenolate sodium [MPS], or azathioprine [Aza]); (2) triple associations with Tac or CsA with Pred and sirolimus; and (3) double associations with Tac or CsA with Pred. For the purpose of the multivariate analysis, data were grouped as regimens, which were mycophenolate-based, Aza-based, or the sum of all other associations.

Laboratory Analysis

Serum creatinine was determined (22) using a Hitachi 912 (Roche Diagnostic System, Basel, Switzerland) isotope dilution mass spectrometry traceable instrument. A freshly voided midstream urine sample was used for the urinalysis and urine culture, and pyuria was defined as a urinary leukocyte count >10⁴/ml. E. coli was isolated in selective culture medium, and antimicrobial susceptibilities were determined by a disk diffusion method (23).

Bacterial DNA Extraction and PCR Conditions

The whole volume of the urine specimen was harvested by centrifugation for 10 minutes at 12,000 rpm in an Eppendorf centrifuge 5810R (Eppendorf, Hamburg, Germany) for the bacterial genomic DNA extraction (24). PCR with specific primers for bacterial 16S rRNA (25) provided the validation of the DNA extraction. The presence of virulence factors was assessed through PCR using separate specific primer pairs for papC and aerobactin (26), papGII and papGIII (27), and M- and G-fimbriae (28); under conditions described by Le Bouguenec et al. (29) for papC and aerobactin; or under conditions described by Johnson et al. (28) for papGII and papGIII and M- and G-fimbriae. All amplification reactions were performed in iCycler BioRad Thermocycler (Bio-Rad Laboratories), and products were electrophoresed on 2% agarose gels, stained with ethidium bromide, and visualized with an ultraviolet transilluminator and digital capture system EDAS 120 (Eastman Kodak Company).

Statistical Analyses

Statistical analyses were conducted with SAS System for Windows, version 8.02 (SAS Institute, Inc.). Categorical variables were compared between groups using χ² or Fisher's exact tests when appropriate. Continuous variables were compared between groups using the Mann–Whitney test. Multiple logistic regression analysis was used to determine host and E. coli virulence factors associated with PN. The level of significance was defined as P < 0.05.

Results

The whole study group consisted of 328 patients, 169 Tx and 159 non-Tx patients, with no statistical differences regarding age (44 versus 41 years). PN was diagnosed in 89 patients, most of them (66.3%) Tx patients (n = 59), whereas Cys occurred in 239 patients, with 46% in Tx patients (n = 110). In the whole sample, women represented most cases of Cys (89.1% versus 10.9%) and PN (76.4% versus 23.6%) compared with men. However, a statistical difference between PN and Cys in the whole sample was observed among men (23.6% versus 10.9%, P < 0.004).

The distribution of the signs, symptoms, and laboratory tests is presented in Table 1. In the Tx patients, women had more Cys (83.6%) and PN (69.5%) than men, but among men, there was a higher frequency of PN than Cys, (30.5% versus 16.4%, P < 0.03). Although Cys was more frequent among non-Tx female than Tx female patients (93.8% versus 83.6%, P < 0.02), the opposite was observed for PN, which was more prevalent among male Tx than non-Tx patients (30.5% versus 10%, P < 0.06). Among PN patients, the presence of fever was significantly different between Tx and non-Tx patients (79.6% versus 100%) because this parameter was an inclusion criterion for the latter group. Pyuria was observed more frequently among PN isolates from Tx than from non-Tx patients (98.3% versus 66.7%). Within the group of Tx patients, the presence of pyuria was also more often observed in PN than in Cys (98.3% versus 85.5%), as opposed to non-Tx patients, in whom pyuria was a more common feature in Cys (87.6% versus 66.7%). Dysuria and urinary frequency were symptoms more commonly observed in Cys from non-Tx than from Tx patients (87.6% versus 64.5% and 33.3% versus 3.6%, respectively).

Table 1.

Signs, symptoms, and laboratory tests

	Tx Patients	Non-Tx Patients	P
PN, n	59	30
gender, n (%)
female	41 (69.5)	27 (90.0)	0.06
male	18 (30.5)	3 (10.0)	0.06
median age (min to max)	42 (18 to 73)	38 (19 to 61)	–
flank pain	–	30 (100.0)	–
fever	47 (79.6)	30 (100.0)	0.007
pyuria	58 (98.3)	20 (66.7)	0.001
increase of serum creatinine	55 (93.2)	–	–
renal allograft tenderness	7 (11.8)	–	–
E. coli-positive hemoculture	6 (10.1)	–	–
renal allograft biopsy	2 (3.4)	–
Cystitis, n	110	129
gender, n (%)
female	92 (83.6)	121 (93.8)	0.02
male	18 (16.4)	8 (6.2)	0.02
median age (min to max)	46 (18 to 69)	38 (18 to 89)	–
dysuria	71 (64.5)	113 (87.6)	0.001
urinary frequency	4 (3.6)	43 (33.3)	0.001
suprapubic pain	7 (6.4)	15 (11.6)	0.1
pyuria	94 (85.5)	113 (87.6)	0.1

n (%), number (percentage) of patients.

The comparisons of host factors, E. coli virulence genes, and antimicrobial resistance in all patients with Cys or PN are listed in Table 2. The multivariate analysis showed significant associations between the occurrence of PN and renal Tx (odds ratio [OR] = 3.12, P < 0.001). Among bacterial factors, significant associations between PN with the presence of papC (OR = 2.03, P < 0.022), the lack of papGIII (OR = 4.88, P < 0.006), and lower rates of resistance to nalidixic acid (OR = 2.90, P < 0.003) were detected. Although papGII was more prevalent in PN than in Cys isolates (27.0% versus 16.7%, P < 0.043), the significance was no longer observed in the multivariate analysis.

Table 2.

The distribution of host and virulence factors of E. coli among all patients with Cys and PN

	Cys (n = 239)	PN (n = 89)	Univariate Analysis			Multivariate Analysis
			OR	95% CI	P	OR	95% CI	P
Host factors
age ≥60 years	40 (16.7)	12 (13.5)	0.78	0.39 to 1.56	0.47
male gender	26 (10.9)	21 (23.6)	2.53	1.34 to 4.78	0.004
previous UTI	104 (43.5)	44 (49.4)	1.27	0.78 to 2.07	0.34
diabetes mellitus	17 (7.1)	10 (11.2)	1.65	0.73 to 3.76	0.23
nephrolithiasis	31 (12.9)	10 (11.2)	0.85	0.40 to 1.81	0.67
PKD	10 (4.2)	2 (2.3)	0.53	0.11 to 2.45	0.41
renal Tx	110 (46.0)	59 (66.3)	2.31	1.39 to 3.83	0.001	3.12	1.80 to 5.39	0.001
Urovirulence genes of E. coli
papC	55 (23.1)	28 (31.5)	1.53	0.89 to 2.62	0.124	2.03	1.11 to 3.72	0.022
papGII	40 (16.7)	24 (27.0)	1.82	1.02 to 3.24	0.043
papGIII	32 (13.4)	4 (4.5)	3.28	1.13 to 9.57	0.029	4.88	1.57 to 15.12	0.006
M-fimbriae	10 (4.2)	3 (3.4)	0.80	0.22 to 2.97	0.74
G-fimbriae	3 (1.3)	0	0.38	0.02 to 7.38	0.29
aerobactin	59 (24.8)	25 (28.1)	1.19	0.69 to 2.05	0.54
Antibiotic resistance
TMP/SMZ	97 (40.6)	41 (46.1)	1.25	0.77 to 2.04	0.37
nalidixic acid	62 (26.0)	12 (13.5)	2.25	1.15 to 4.41	0.018	2.90	1.42 to 5.93	0.003
pipemidic acid	64 (26.8)	14 (15.7)	1.90	1.01 to 3.61	0.049
ciprofloxacin	55 (23.1)	12 (13.5)	0.52	0.26 to 1.02	0.06
norfloxacin	55 (23.1)	12 (13.5)	0.52	0.26 to 1.02	0.06
ampicilin	119 (49.8)	47 (52.8)	1.13	0.69 to 1.84	0.63
cefalotin	60 (25.1)	26 (29.2)	1.23	0.72 to 2.12	0.45
ceftriaxone	5 (2.1)	1 (1.1)	0.53	0.06 to 4.62	0.56
gentamicin	17 (7.1)	4 (4.5)	0.62	0.20 to 1.88	0.39
nitrofurantoin	0	0	–	–	–

n (%), number (percentage) of patients. CI, confidence interval.

Comparisons of host and virulence factors in Tx patients with PN and Cys are presented in Tables 3 and 4. The risk of occurrence of PN was significantly higher among men (OR = 2.35, P < 0.037) and for those no longer receiving TMP/SMZ prophylaxis (OR = 2.73, P < 0.019). The percentage of PN and Cys among patients receiving TMP/SMZ was significantly different (17.0% versus 30.9%). PN was associated with the lack of papGIII (OR = 5.07, P < 0.041) and with lower rates of resistance to pipemidic acid (OR = 2.57, P < 0.014). There were statistical differences between PN and Cys isolates with respect to papGII (23.7% versus 11.0%, P < 0.03) and for resistance to nalidixic acid (20.3% versus 37.3%, P < 0.03). However, such differences were no longer observed in the multivariate analysis. Nitrofurantoin resistance was not identified in any isolated strain in the study presented here.

Table 4.

The distribution of E. coli virulence factors among Tx patients with Cys and PN

	Cys (n = 110)	PN (n = 59)	Univariate Analysis			Multivariate Analysis
			OR	95% CI	P	OR	95% CI	P
Urovirulence genes of E. coli
papC	18 (16.5)	14 (23.7)	1.59	0.73 to 3.48	0.25
papGII	12 (11.0)	14 (23.7)	2.54	1.09 to 5.93	0.03
papGIII	15 (13.6)	2 (3.4)	4.50	0.99 to 20.38	0.05	5.07	1.07 to 24.13	0.041
M-fimbriae	5 (4.6)	1 (1.7)	0.36	0.04 to 3.17	0.36
G-fimbriae	2 (1.8)	0	0.36	0.02 to 7.72	0.30
aerobactin	24 (21.8)	14 (23.7)	1.12	0.53 to 2.36	0.78
Antibiotic resistance
TMP/SMZ	57 (51.8)	30 (50.9)	0.96	0.51 to 1.81	0.90
nalidixic acid	41 (37.3)	12 (20.3)	2.33	1.11 to 4.89	0.03
pipemidic acid	46 (41.8)	13 (22.0)	2.54	1.23 to 5.24	0.01	2.57	1.21 to 5.46	0.014
ciprofloxacin	37 (33.6)	12 (20.3)	0.50	0.24 to 1.05	0.07
norfloxacin	37 (33.6)	12 (20.3)	0.50	0.24 to 1.05	0.07
ampicilin	64 (58.2)	33 (55.9)	0.91	0.48 to 1.73	0.78
cefalotin	39 (35.4)	21 (35.6)	1.01	0.52 to 1.95	0.98
ceftriaxone	2 (1.8)	1 (1.7)	0.93	0.08 to 10.49	0.95
gentamicin	11 (10.0)	4 (6.8)	0.66	0.20 to 2.15	0.48
nitrofurantoin	0	0	–	–

n (%), number (percentage) of patients.

Table 3.

The distribution of host factors among Tx patients with Cys and PN

	Cys (n = 110)	PN (n = 59)	Univariate Analysis			Multivariate Analysis
			OR	95% CI	P	OR	95% CI	P
Host factors
age ≥60 years	19 (17.3)	9 (15.3)	0.86	0.36 to 2.05	0.74
male gender	18 (16.4)	18 (30.5)	2.24	1.06 to 4.75	0.04	2.35	1.06 to 5.22	0.037
previous UTI by E. coli	60 (54.5)	33 (56.0)	1.06	0.56 to 2.00	0.86
diabetes mellitus	15 (13.6)	10 (17.0)	1.29	0.54 to 3.09	0.56
nephrolithiasis	3 (2.7)	1 (1.7)	0.62	0.06 to 6.05	0.68
PKD	9 (8.2)	2 (3.4)	0.39	0.08 to 1.89	0.24
VUR	2 (1.8)	1 (1.7)	0.93	0.08 to 10.49	0.95
deceased donor (versus living)	53 (48.2)	29 (49.2)	1.04	0.55 to 1.96	0.90
Gregoir versus Politano anastomosis	75 (68.2)	40 (67.8)	1.02	0.52 to 2.00	0.96
TMP/SMZ prophylaxis	34 (30.9)	10 (17.0)	2.19	0.99 to 4.84	0.05	2.73	1.18 to 6.29	0.019
use of double J stent	23 (20.9)	19 (32.2)	1.80	0.88 to 3.67	0.11
Immunosuppression regimen
mycophenolate-baseda	43 (39.1)	21 (35.6)	1.00	–	–
Aza-based	37 (33.6)	25 (42.4)	1.38	0.67 to 2.87	0.38
others	30 (27.3)	13 (22.0)	0.89	0.39 to 2.04	0.78

n(%), number (percentage) of patients.

^aReference.

When we categorized Tx patients according to the time after transplantation (data not shown in the table), we did not identify a significant increase in the occurrence of PN per each 6-month period (OR = 1.03, 95% confidence interval 0.990 to 1.077, P < 0.135).

There were no statistical differences between the percentage of patients with Cys or PN regarding immunosuppressive induction with basiliximab (20.0% versus 18.6%), daclizumab (1.8% versus 5.1%), alemtuzumab (1.0% versus 0.0%), muromonab-CD3 (1.0% versus 0.0%) and thymoglobulin (7.2% versus 6.8%). The various maintenance immunosuppressive regimens were also not statistically different with respect to Cys or PN isolates from patients receiving Tac/Pred/MMF (5.5% versus 3.4%), Tac/Pred/Aza (19.1% versus 16.9%), CsA/Pred/Aza (14.5% versus 25.4%), CsA/Pred/MMF (3.6% versus 5.1%), CsA/Pred (6.3% versus 3.4%), Tac/Pred/MPS (20.1% versus 16.9%), Tac/Pred (3.6% versus 1.7%), CsA/Pred/MPS (10% versus 10.2%), or Siro/Pred/Tac (6.4% versus 1.7%).

Discussion

The severity of a UTI is determined by the virulence of the infecting strain modulated by the innate immune response of the host (30). Although there are several studies assessing E. coli virulence genes in upper UTI (5,11–13), only one of them has been conducted in renal Tx recipients (21). We aimed to evaluate host and bacterial virulence factors mostly associated with PN by E. coli in Tx patients.

In the study presented here, renal Tx was associated with a 3-fold increase in the risk of developing PN (OR = 3.12). Such a finding raises the hypothesis that immunosuppression, neo ureterovesical implantation, shorter ureter, manipulated bladders, and other conditions inherent in renal Tx may contribute to a higher risk of PN. However, we did not find statistical differences among different immunosuppression schemes, namely triple-drug regimens using mycophenolate or Aza and others (sirolimus-based or double-drug regimens) with respect to the occurrence of PN, as also observed by Pellé et al. (6). Notwithstanding, Kamath et al. (31) did find an association with the use of Mycophelonate and PN in Tx patients, but the number of patients was smaller than ours.

Some studies reported a greater prevalence of women (6,32), whereas others did not find any sex-related differences in the frequency of posttransplantation PN (31). Nevertheless, these studies (31,32) differ from ours because they did not compare PN with Cys, but rather patients who developed PN versus those who never did. In the study presented here, although any UTI event by E. coli was statistically more frequent in women than in men, when we compared PN with Cys patients, it became evident that PN in Tx patients occurred more often in men, and the risk of PN was 2-fold higher in men. Although other investigators also observed higher rates of PN among men (11), the reasons for the predilection of male gender in the series of Tx patients presented here remain unclear. Our findings could not be ascribed solely to older age and consequently to prostatism because the Cys and PN groups were young and comparable with respect to age (49.3 ± 13.1 years versus 46.5 ± 14.8 years, respectively NS). In a recent retrospective analysis based on data from the U.S. Renal Data System, Hurst et al. (33), have shown that benign prostatic hyperplasia (BPH) is very common in men after renal Tx and is independently associated with UTI. Age was independently associated with BPH in their reports, but whether BPH was causing more upper (PN) than lower (Cys) UTI could not be further specified (33). One potential limitation of this study is that it was not possible to entirely rule out the effect of BPH or prostatism in the series presented here because of its retrospective design. Finally, according to Sadeghi et al. (34), male renal Tx recipients with UTI might present a stronger inflammatory cytokine response than female patients who block inflammatory responses by producing soluble IL-1 receptor antagonist, possibly because of continuous stimulation of the bladder by insignificant bacteriuria. Whether this stronger response may be linked to the invasion of the upper urinary tract, especially in Tx patients, remains to be determined.

Specific causes of ESRD (e.g., diabetes mellitus, autosomal dominant PKD, nephrolithiasis, or VUR) were not associated with an increased risk of PN in the study presented here, in agreement with other reports (6,31). According to Rice et al. (21), diabetes mellitus was even more associated with Cys rather than with PN among Tx patients. It is possible that the lack of association between PN with autosomal dominant PKD, nephrolithiasis, and VUR with native kidneys was because of the small number of these diseases in the sample presented here.

We did not find any association between PN and the use of a double J stent, in accordance with the study of Kumar et al. (35). On the other hand, Kamath et al. (31) observed a 4-fold increase in the risk of PN in Tx patients with ureteric stents. These findings might have been accounted for by an earlier removal of stents in our service (36) compared with others (31).

Kidneys from deceased donors and Gregoir-type of anastomosis did not show higher rates of PN in the sample presented here, in agreement with other studies (6,31,32).

In this study, the percentage of patients receiving TMP/SMZ prophylaxis against Pneumocystis carinii was significantly lower in the PN than in the Cys group, and the multivariate analysis showed a risk almost 3-fold lower of developing PN when on TMP/SMZ (OR = 2.73), despite high TMP/SMZ resistance rates (approximately 50% in both groups). A separate analysis of resistance to TMP/SMZ only on patients who developed either PN or Cys receiving TMP/SMZ prophylaxis showed a nonsignificant trend for higher rates of resistance in PN (8 of 10 [80%]) versus Cys (21 of 34 [62%]). Other reports also observed decreased rates of UTI associated with TMP/SMZ prophylaxis (37,38) but not particularly with PN.

There is no conclusive evidence whether or not the presence of virulence factors is correlated with differences in susceptibility to antimicrobials. Although some investigators suggested that virulence genes increased antibiotic resistance of previously resistant strains (16), others observed that quinolone-resistant E. coli strains were more prone to induce Cys rather than PN (17,18,20) because of decreased renal invasive capacity concomitantly acquired by the mutation. Our findings, in agreement with those reports, showed a lower proportion of E. coli resistance to first-generation quinolones in PN isolates: the risk of resistance in PN isolates to nalidixic acid was reduced in the whole sample (OR = 2.90) and to pipemidic acid in the Tx group (OR = 2.57). Resistance to ciprofloxacin or norfloxacin was also less frequent in PN isolates, almost achieving statistical significance. We did not observe a single case of fluorquinolone resistance among the PN isolates of non-Tx patients. We hypothesize that the slightly higher resistance to fluorquinolones in PN isolates from Tx patients compared with the non-Tx patients might have been consequent to their more frequent use in this population, selecting resistant strains.

The reported prevalence of papC-positive E. coli strains isolated in patients with UTIs is highly variable—10% to 77% (5,13,28,39–41), and only two studies have found a higher prevalence of papC-positive E. coli in PN versus Cys isolates (50% versus 10% (13) and 76% versus 35% (5)). These studies did not include patients with renal Tx. The current prevalence of papC in PN isolates from Tx patients was not significantly different from Cys isolates. Although the multivariate analysis suggested risk for papC in PN isolates in the series presented here (OR = 2.03), it was not an independent risk given that the univariate analysis was NS.

Previous studies have demonstrated that among the three classes of papG genes, only papGII is associated with PN (5,11,12,15). In the study presented here, the univariate analysis showed papGII to be more often associated with PN than with Cys in the whole sample and also in Tx patients. However, it did not represent an independent risk factor for PN. Our prevalence of 27% of papGII in PN is not high if compared with other studies that reported rates of 54% to 93% in the non-Tx population (5,11,12,15). In Tx patients, Rice et al. (21) found significant differences regarding the papGII prevalence between PN and lower UTIs—47% versus 18%. Nevertheless, in experimental models, Mobley et al. (42) have shown that adherence mediated by the P-fimbrial adhesin played only a subtle role in the development of PN. Additionally, it has been observed that although PapGII enhanced the establishment of kidney infections by E. coli in mice, infections did not persist because of the immune response (43). According to Tseng et al. (11), P-fimbria virulence factors have been suggested to be less common under immunosuppression, although the exact distribution of papGII among isolates of PN from immunosuppressed patients was not specified (11). Some investigators observed a lower predominance of papGII in E. coli isolated from UTI patients with urinary abnormalities, either in children (44,45) or in adults (46). Although patients with neurogenic bladder were excluded from the sample presented here, UTI occurring in a transplanted patient is already considered as a complicated one (47) because of the heterotopic position of the graft and the shorter ureteral length. Finally, studies that observed a high prevalence of papGII in PN could have been biased by the exclusion of quinolone-resistant strains (5) because the latter would possibly possess fewer virulence factors (5,13).

Several studies have demonstrated that PapGIII is mostly identified in acute Cys strains (48–50) because Forsmann antigen, the host cell receptor, is rare in renal cells. This is in agreement with our observation that the papGIII prevalence was higher in Cys versus PN in all patients including the Tx ones. Thus, the lack of papGIII in E. coli strains was an independent factor associated with PN in all patients (OR = 4.88) and Tx patients (OR = 5.07). On the other hand, some studies did not find any association between papGIII and PN in non-Tx (5,11,13,15) or Tx patients (21), with the prevalence ranging from 4% to 35%.

The aerobactin prevalence in E. coli strains is also highly variable in the literature, from 26% to 85% (5,11,13,15,28,39,40). In this study, the prevalence of aerobactin was up to 28% but not significantly different between the PN and Cys isolates, coinciding with some reports (11,15) but not with others (5,13). There are few studies about the association of M- and G-fimbriae with UTIs (5,28,41) and none in a Tx setting. The current prevalence of these fimbriae was very low (up to 4.2%, in agreement with other authors (28,41)) and did not differentiate PN from Cys.

In conclusion, the data presented here suggest that renal Tx increased the risk for PN, and the male sex represented a host factor independently associated with risk, whereas the prophylaxis with TMP/SMZ was protective. Although papGII was more prevalent in PN than in Cys, it was not independently associated with PN in Tx, whereas the lack of papGIII was considered as an independent risk factor for PN. Immunosuppression, abnormal urinary tract, and resistance to quinolones are some features that might have explained the lower prevalence of papGII in PN, suggesting that bacteria presenting with fewer virulence genes may invade the upper urinary tract in Tx patients.

Another limitation of this and other studies is that only the prevalence rather than expression of virulence factor genes have been evaluated. Studies examining the expression of such virulence factors in lower and upper UTIs in renal Tx recipients must be conducted to further confirm such findings.

Disclosures

None.