To investigate skeletal growth and bone metabolism in a chronic animal model of urinary diversion.
To investigate skeletal growth and bone metabolism in a chronic animal model of urinary diversion.
Young male Wistar rats (120) were allocated randomly to four groups undergoing: ileocystoplasty, ileocystoplasty and resection of the ileocaecal segment, colocystoplasty, and controls. All animals received antibiotics for 1 week after surgery; half of each group remained on oral antibiotics. Bone-related biochemistry was measured in serum and urine. Dual-energy X-ray absorptiometry and peripheral quantitative computed tomography (pQCT) were used to determine bone mass ex vivo.
Most (90%) of the rats survived the study period (8 months); six rats died from bowel obstruction at the level of the entero-anastomosis and four had to be killed because of persistent severe diarrhoea. Vital intestinal mucosa was found in all augmented bladders. There were no differences in bone length and volume. Loss of bone mass was almost exclusively in rats with ileocystoplasty and resection of the ileocaecal segment (−37.5%, pQCT, P < 0.01). There was no hyperchloraemic metabolic acidosis or gross impairment of renal function. Hypomagnesaemia, hypocalcaemia and decreased insulin-like growth factor-binding protein 3 were the only significant findings on blood analysis. Deoxypyridinoline crosslinks in urine were higher in rats with an enterocystoplasty than in controls.
Enterocystoplasty in rats neither impairs skeletal growth nor bone quantity, but leads to significant loss of bone mass when combined with resection of the ileocaecal segment. Rarefaction of the trabecular network is confined to the metabolically highly active cancellous compartment, most likely as a consequence of intestinal malabsorption.
Almost 80 years after Simon of London had for the first time ‘directed the orifices of the ureters into the rectum’ (with only temporary success) in 1852 , his fellow-countryman Turner incidentally described generalized rickets following a similar procedure in a patient with bladder exstrophy . Boyd  finally proposed a causal link between ureterosigmoidostomy, chronic acidosis and metabolic bone disease. When continent intestinal urinary reservoirs gradually gained popularity, the issue was revived in a series of mostly retrospective clinical investigations [4–14]. At the same time some authors also postulated delayed linear growth [4,15,16] and development [17,18] in children and adolescents undergoing such surgery. However, the groups studied were heterogeneous, with numerous confounding factors and often conflicting results . Despite these efforts most questions, especially about the mechanisms of bone loss and growth failure, remained largely unanswered. Given the complexity of the subject it seemed justified to undertake an animal experiment allowing a more systematic approach.
All experiments were approved by The UK Home Office. The study comprised 8–10-week-old male Wistar rats (170–220 g; Harlan Orlac, Bicester, Oxon, UK), and experiments were delayed for at least 1 week after delivery. The animals were housed one per cage after surgery in controlled atmospheric conditions (20–21°C; 55% humidity; 12-h cycles of light and darkness, 08.00–20.00 hours). Except for the day before surgery the animals had continuous and free access to standard rat chow and water.
On the day of surgery the animals were randomly assigned (30 each) to one of four different procedures. Anaesthesia was administered intraperitoneally (Vetalar®, ketamine hydrochloride, 100 mg/mL for injection, Parke-Davis Veterinary, Pontypool, UK; Rompun®, xylazine, 2% solution for injection, Bayer UK Ltd., Bury St. Edmunds, UK). The animals were placed supine on a heated operating table. After the abdomen had been shaved and disinfected with alcohol a full midline incision was made. Control rats (group 0) underwent sagittal cystotomy and closure only (running 7/0 chromic collagen suture); in group 1 (ileum, ileocystoplasty) a well-vascularized 1.5 cm length of ileum was selected ≈ 7 cm proximal to the ileocaecal valve, isolated and opened on its antimesenteric border. The patch was sutured to the bladder without twisting or tension using an operating microscope (running 7/0 chromic or 8/0 chromic collagen sutures); in group 2 (ileum + resection) ileocystoplasty was combined with resection of the ileocaecal segment; in group 3 (colon, colocystoplasty) a 10-mm length of sigmoid colon immediately adjacent to the urinary bladder was isolated. After detubularization, enterocystoplasty was completed as described. In groups 1–3 the bowel was re-anastomosed microsurgically after spatulation with interrupted inverting 8/0 silk sutures. The abdominal wall was closed with running 4/0 chromic fascial and skin sutures. Animals which died before the end of the study were examined as soon as possible.
At 1 h before surgery all rats received 10 mg/kg enrofloxacin (Baytril®, 5% solution for injection, Bayer). Oral antibiotic therapy (Baytril®, 10% oral solution, 50 mg/L) was supplied via the drinking water for the first week after surgery in all animals. Half of the rats in each group were randomly chosen to remain on oral antibiotic prophylaxis (AB).
Each animal was weighed on delivery, immediately before surgery and weekly thereafter until the end of the study. Six rats in each group were randomly chosen for an estimation of deoxypyridinoline crosslinks (DPDX, a bone resorption marker) in urine 3, 6, 12, 20 and 28 weeks after surgery. For 16 h the animals were housed in a metabolic cage. Blood gases were analysed 1 day before the animals were killed.
As bone disease was assumed to be a long-term consequence of urinary diversion, the duration of the study was 8 months, after which all surviving animals were killed by direct cardiac exsanguination, after thoracotomy, under general anaesthesia. Blood was collected between 10.00 and 14.00 hours. In animals of groups 1–3 the bladder was exposed and transected at the level of the bladder neck. Orthotopic bowel (corresponding to the origin of the patch), spleen, both kidneys, and the right lobe of the liver were excised, fixed in 4% formaldehyde solution and stored at 4°C.
After preparing the hip-joints both lower extremities were disarticulated. The lower thoracic and complete lumbar spine was collected. The right tibiae were fixed in 4% formaldehyde solution in preparation for conventional histology. The left tibiae and left femora were fixed in 40% alcohol for 24 h, dehydrated through graded alcohols (70%) in the following week and stored at 4°C. The right femora and lumbar spine were stored in 0.9% saline at − 20°C.
Biochemical variables relating to renal, liver and bone function were measured using standard colorimetric assays using a multichannel autoanalyser (Dax-48, Bayer Diagnostics, Basingstoke, UK). The biologically active intact and N-terminal forms of parathyroid hormone (PTH) were measured using the rat PTH immunoradiometric assay (IRMA) kit (Nichols Institute Diagnostics, San Juan Capistrano, CA, USA). 25-Hydroxycalciferol levels in serum were determined using a 25-hydroxyvitamin D 125I radioimmunoassay (RIA) kit (INCSTAR Corp., Stillwater, Minnesota, USA). IGF-I was measured using the IGF-I IRMA kit (100T, Nichols). IGF-II and IGF-binding protein-3 (IGFBP-3) were determined using an IRMA (DSL-2600 Active® and DSL-6600 Coated Tube Kit, both Diagnostic Systems Laboratories Inc., Webster, TX, USA).
DPDX excretion in urine was measured using the Pyrilinks®-D immunoassay (Metra Biosystems, Mountain View, CA, USA).
Macroscopically abnormal and representative sections of all specimens (kidneys, urinary bladder) were prepared for routine histological processing. The lengths of both femora and the left tibia were measured with callipers from the top of the greater trochanter to the end of the medial condyle, and from the intercondylar bar to the medial malleolus, respectively. The volume was estimated by weighing the bone specimens before and during immersion in water.
Longitudinal sections of the right tibia were processed according to Masson-Goldner and examined for mineralization, thickness of osteoid seams, resorption surfaces, number of osteoclasts and osteoblasts, marrow fibrosis and density of the trabecular network.
DEXA was used to measure bone mineral content (BMC) and density (BMD). Scans of lumbar vertebral bodies L2 to L5, the left femur and the left tibia were taken ex vivo by DEXA (resolution 0.5 × 0.5 mm pixel size; scan speed 30 mm/s) on a Norland XR-36 scanner.
pQCT is a sensitive and reliable noninvasive tool for assessing changes in bone mass and architecture in small animals, and can separate the cancellous and the cortical compartments by using algorithms, and predict mechanical properties . Measurements were made in the proximal metaphysis of the left tibia at 4.5 mm from the knee joint on a tomograph (XCT-960, Stratec, Pforzheim, Germany). A slice thickness of 1.2 mm and the smallest possible voxel size of 0.197 × 0.197 × 1 mm was chosen. Periosteal and endocortical perimeter, polar moment of inertia and resistance were either measured or calculated.
For all independent variables (weight, serum tests, DPDX) a Kruskal-Wallis test was used as a nonparametric anova, with differences considered significant at P < 0.05. The results are presented as the mean of Wilcoxon rank sums. In the longitudinal measurements of DPDX levels the Krauth test was applied to compare time series of repeated measurements (dependent variables).
Other data (bone dimensions, DEXA, pQCT) were assessed by using one-way anova, with equality of variances tested using the Levene F-test, and Bonferroni-adjusted significances determined for multiple pair-wise comparisons using pooled-variance t or separate variance t-tests for comparisons with equal or unequal variances, respectively. All statistical tests were two-tailed. Data were analysed by comparing the mean values for each variable of the enterocystoplasty groups with the corresponding variable of the control group.
All animals survived the surgery and the first 2 days afterward; 108 rats (all controls, 25, 26 and 27 in groups 1–3, respectively) survived the entire 8 months. Six rats died from bowel obstruction at the level of the entero-anastomosis, and four were killed because of persistent severe diarrhoea; in two animals the cause of death was unknown.
There was no difference in weight gain and final weight in group 3 compared with the controls; the weight gain and final weight in groups 1 and 2 were significantly lower than in controls (P < 0.004).
There were no significant differences (pH, pCO2, bicarbonate, serum chloride) between controls and groups 1–3 in blood gas analysis. There was significant hypomagnesaemia (P < 0.002), hypocalcaemia (P < 0.03) and reduction of IGFBPs in group 2 (P < 0.01; Table 1). When the duration of antibiotic treatment was considered, alkaline phosphatase was significantly greater in group 2 (no AB) and lower in group 2 (AB).
There were no significant differences among the groups in bone-related serum biochemistry and renal function, as grossly indicated by normal creatinine. There were significant differences only at 5 and 7 months in DPDX levels among the groups, with the lowest values in the control (P < 0.029 and P < 0.008) (Fig. 1). The DPDX levels with time within each group were significantly different for each group, the values being highest initially and then continuously decreasing (all P < 0.001). In the course of DPDX excretion over time (time series) there was a significant difference only between the controls and groups 1–3 together (P < 0.047). The duration of antibiotic medication had no effect on DPDX levels in rat urine.
Only 41 animals had histologically normal kidneys, with pathological changes in all groups. Vital intestinal mucosa was found in all enterocystoplasty specimens, similar to that in corresponding orthotopic bowel segments from the same animal (Fig. 2). There were no significant differences in bone length and volume (Table 2). The long bones showed a normal macroscopic anatomy with epi-, dia-, and metaphysis. There was neither abundant osteoid nor signs of active resorption. In group 2 the trabecular network was distinctly rarefied. Cortical bone was normal in all animals.
|Mean variable||Group, no AB||Group, AB|
|Mean (sd) pQCT values|
|Cross-sectional||702.8 (9.1)||656.3 (25.4)||609.4 (26.9)||683.6 (12.1)||688.7 (15.3)||671.7 (15.3)||657.2 (12.8)||692.0 (13.6)|
|Cortical||1271 (8.5)||1243 (15.8)||1252 (15.1)||1256 (9.5)||1266 (11.6)||1263 (10.9)||1264 (15.9)||1273 (8.3)|
|Cancellous||247.4 (13.4)||222.1 (14.8)||154.6 (11.7)||242.5 (14.3)||237.7 (21.0)||227.7 (19.4)||180.7 (10.3)||230.3 (9.4)|
There were no differences in BMC and BMD in rats in group 3 and controls. In group 2 (no AB) the BMC and BMD were significantly reduced in lumbar vertebrae (−15.1%, −17.3%), femora (−13.2%, − 10.2%) and tibiae (−11.8%, −9.3%) (P < 0.01). The effect was either sustained (vertebrae; −10.9%, −12.32%; P < 0.01) or attenuated on long-term antibiotics (femora, tibiae; P < 0.1). The attenuation was not significant. There was also a mild decrease in BMD in group 1 (−5.1%; P < 0.1; Table 2).
On pQCT, there were no differences in bone geometry (area, perimeter), apart from reduced cortical area in group 2 (−9.3%, P < 0.05). In the same group there were significant reductions in cross-sectional, cortical and cancellous BMC (−16.3%, P < 0.01; − 11.0%, P < 0.05; − 37.9%, P < 0.01) and a dramatic decrease in cancellous BMD (true volumetric density; − 37.5%, P < 0.01), but cortical BMD was no different from controls; the effect was attenuated by long-term antibiotics (−24.0%) but not significantly (Table 2).
As bowel was not meant to serve as a storage vehicle for urine, the use to which it is put in urology may result in numerous complications in the short- and the long-term [18,19]. The potential complications are a result of either reduction of the absorptive bowel capacity through functional loss of those segments required for reservoir construction, or the highly unphysiological contact of the reconfigured bowel with urine. It is therefore critical for the validity of any animal model of urinary diversion to have vital intestinal mucosa rather than fibrotic transformation of the bowel segments after incorporation into the urinary tract using microsurgical techniques (Fig. 2).
A potential long-term effect of particular concern with urinary diversion is bone demineralization, which was reported in children and adults in the first half of the last century [2,3]. Apart from poor intestinal absorption of calcium and vitamin D, chronic acidosis has been consistently blamed for the main skeletal effects. It is thought to act through direct physicochemical noncell-mediated cation-hydrogen exchange at the bone surface (bone as a buffer system), stimulation of osteoclastic bone resorption and inhibition of osteoblastic bone formation [21–23]. However, in the clinical reality of modern continent diversion the role of acidosis and bone disease is less clear. In none of the available contemporary case series was overt derangement of the acid-base-balance, rickets or osteomalacia encountered [5–14]. Bone demineralization was found in only three series, in all instances minor and asymptomatic [5,8,10]. Despite equally sophisticated methods and a follow-up of up to 30 years, the other series failed to show such changes, even in the presence of acidosis [4,6,7,9,11–14].
The rat model has been well established for studying bone quality in preclinical osteoporosis research . All methods used to assess bone quality and metabolism in this study were standardized for small animals and correspond appropriately to those applied in humans. However, in urinary diversion, animal experiments are far from being standardized. More than 17 different dog and rat models have been used to investigate metabolic complications of such surgery. While rat and human skeletons have many features in common, there are distinct differences in the anatomy and physiology of the gastrointestinal and urinary tract between these species. A different ratio of excluded bowel length to bladder capacity and a shorter urine contact time in rats are among the numerous factors which clearly limit the applicability of the present and previous reports to the analogous human situation.
As the renal capacity to excrete acid loads is many times greater in rats than in man, while maintaining a normal or near-normal serum pH, urinary diversion alone usually does not create detectable acidosis in these animals [25–27]. Only McDougal et al. found a minimal systemic acidosis in their rats, but without providing exact data or even significance levels; their observation that the administration of bicarbonate and ascorbic acid prevents a reduction of calcium content in bone-ash studies, and decrease of bone density in dual-photon absorptiometry (DPA), respectively (in an unknown proportion of animals), is usually used to support the assumption that even a minimal disturbance of the acid-base-balance may cause bone disease. Similar findings were published by Hochstetler et al., who stressed their augmented animals with ammonium chloride to mimic the human condition; when rats with augmentation cystoplasty given 1% ammonium chloride were supplemented with an equal molar diet of sodium bicarbonate, metabolic acidosis resolved and bone mineral density normalized to control values (with an unexplained discrepancy of bone mineral ash content and DEXA). Interestingly, Roth and Gasser  found no bone pathology in rats with respiratory compensated acidosis using up-to-date technology (DEXA, histomorphometry). The present rats had neither acidosis nor hyperchloraemia, suggesting that all changes in bone metabolism were caused by other factors.
This is the first study with the complementary use of DEXA, pQCT, conventional bone histology, histomorphometry, bone ashing and biomechanical testing in the evaluation of bone mass and structure . McDougal et al. used whole-body measurements in 30 of originally 78 rats (mortality and selection criteria not given) using DPA; interestingly, DPA was not applied in rats with bicarbonate supplement in the drinking water, the group on which their argument was mainly based. Hochstetler et al. were the first to use serial whole-body measurements (in vivo, DEXA); although not significantly different at any one time, regression analysis identified significant differences in skeletal mineralization. Roth and Gasser  found no differences in BMD (by DPA) when comparing acidotic rats, rats with enterocystoplasty and controls. The major disadvantage of DPA and DEXA is their inability to distinguish between spongiosa and cortex. In contrast, pQCT can be used at any peripheral site, and changes in cancellous and cortical bone can be monitored separately and noninvasively over time, providing a true volumetric BMD .
All the methods in the present study conclusively showed bone changes limited to the highly active cancellous compartment through rarefaction of the trabecular network, reducing the endosteal bone surface and expanding the marrow cavity area. This pattern is typical of metabolic bone disease and resembles that of osteoporosis. This is in contrast to the findings by Roth and Gasser , who reported endocortical bone degradation. However, in all instances a mineralization defect (osteomalacia) and renal osteodystrophy could be excluded.
Normal bone mineral metabolism requires the interaction of calcium, phosphate and magnesium, which are influenced by PTH, calcitonin and vitamin D. Although these variables should be accountable in all discussions of bone demineralization, they have been measured in only two further animal studies [26,30]. Normal PTH and vitamin D levels in all clinical and experimental studies, including the present, mean that any contribution of these hormones to bone disease in continent urinary reconstruction is extremely unlikely.
Significant reductions of serum calcium and magnesium in the present rats were fully or partly consistent with the reports by McDougal et al. and Roth and Gasser  (hypocalcaemia only), while others [25,27,30] found no differences in serum electrolytes. Perhaps hypocalcaemia (with normal albumin) in the present study was a consequence of hypomagnesaemia, as observed in PTH-resistant hypoparathyroidism [31,32]. Despite a low normal phosphate, calciopenic rickets (secondary hyperparathyroidism) is unlikely with normal alkaline phosphatase and PTH. Severe hypomagnesaemia in the present rats was most probably a results of malabsorption . Malabsorption as a causal factor is supported by the observation of loose stool and a significant reduction of final weight and IGFBP-3, an indicator of intestinal failure , in animals with magnesium deficiency.
IGF-I and II are related growth hormone-dependent peptide factors which mediate many of the anabolic and mitogenic actions of growth hormones. The IGFs circulate in plasma complexed to a family of BPs which either inhibit (IGFBP-1, -2 and -4) or stimulate (IGFBP-3 and -5) IGF action. Normal IGF-I levels usually exclude growth hormone deficiency [34,35]. As the IGF system is important in regulating bone growth and formation, in the present study IGF-I, -II and the clinically most valuable BP-3 (stimulatory component) were determined for the first time in a model of urinary diversion. Not surprisingly, IGF-II could not be detected in these rats, as in rats and mice IGF-II levels are highest in the fetus and decline rapidly after birth to low levels in the adult.
The significant reduction of IGFBP-3, which is mainly regulated by nutritional status, age and growth hormone, in the present rats was probably caused by malabsorption . Barksdale et al. identified low IGFBP-3 as index of intestinal failure in children with short-bowel syndrome, resulting in growth failure, by anthropometry. Although the link between IGFBP-3 reduction and bone loss is unclear, Boonen et al. found a down-regulation of stimulatory components of the IGF system in the pathogenesis of age-related (type II) femoral-neck osteoporosis, leading to impaired bone formation. However, treating ovariectomized osteopenic rats with the complex of rhIGF-I/IGFBP-3 increased bone mass and improved structure in the femoral neck .
DPD is a mature amino acid, forming covalent cross-links between adjacent collagen chains in extracellular matrix . It is not susceptible to dietary influences and is found in significant amounts only in bone and collagen. Its measurement allows the quantification of bone resorption. Given the few animals assessed, serial DPD measurements must be interpreted with caution. In all groups bone resorption was highest immediately after surgery and decreased over time, suggesting that it is a general effect related to surgical trauma rather than to a particular procedure. A similar course was reported in the only clinical study of DPD in 46 men after three different types of urinary diversion ; urinary DPD reached the highest levels soon after surgery, gradually decreasing to a stable level within 1–2 years. However, interestingly, at the end of the present study, DPD was significantly lower in the control group and comparison of time series of DPD showed the same tendency. The overall results suggest a decreasing bone turnover with a relatively increased resorptive activity after enterocystoplasty.
The effect of functional bowel loss in continent urinary reconstruction is mainly related to bowel type and length . While it cannot be assumed that the physiology of the ileocaecal valve is similar in rats and man, this structure is similarly important in both species. In humans, the terminal ileum and ileocaecal valve are particularly critical for resorptive function, regulation of bowel emptying and compensation of small bowel loss; resection of up to 60 cm ileum is not thought to result in malabsorptive sequelae if the terminal ileum and ileocaecal valve are intact . In rodents, the ileocaecal segment is a vital structure for the digestion of cellulose, the primary part of their diet, particularly in a laboratory setting. As most of the dramatic changes occurred in animals with ileocystoplasty and resection of the ileocaecal valve, the latter deserves particular attention. In patients with continent urinary reservoirs Roth et al. estimated the risk of bowel dysfunction with chronic diarrhoea after resection of an ileocaecal segment to be twice as high as after ileal resection. When Simmons et al. studied the effect of small bowel resection or bypass on the rat skeleton they found increasing osteopenia 1–5 weeks after surgery. A review of previous reports support intestinal malabsorption/maldigestion as a major factor in the loss of bone in the present rat model.
The decision to assess groups with short- or long-term antibiotic treatment was based on a clinical study suggesting that the prophylactic administration of such drugs can prevent growth impairment after urinary diversion . As chronic bacteriuria is common in patients with continent urinary diversion, and raised serum antibody titres against Escherichia coli and Proteus mirabilis have been detected in patients with conduit diversion , a direct or cytokine-mediated effect on bone turnover may be possible. However, antibiotics can influence bone formation and resorption independently of their anti-infectious properties . Although fluoroquinolones may adversely affect cartilage growth and endochondral ossification in children, long-term enrofloxacin in the present rats tended to limit bone loss, but not statistically significantly. As the water intake was not measured, the exact doses of antibiotics are unknown.
The assumption that urinary diversion through intestinal segments in childhood has a detrimental effect on skeletal growth is almost exclusively based on three retrospective clinical series with numerous methodological limitations [4,15,16]. In a more recent study of 123 children with enterocystoplasty and a mean follow-up of> 8 years , the authors concluded that the loss of percentile position in 15% of patients is a general effect that must be considered in any clinical population of the same size and age distribution after the same length of time. In the children with the longest follow-up, most caught up with their preoperative position and achieved their genetic growth potential. In the present study, age at surgery was chosen to avoid the extremely high per- and postoperative mortality in even younger rats . The ensuing disadvantage that the normal growth period of rats (21–120 days) was only partly covered is outweighed by the fact that epiphyses in rats do not close before the age of 12–24 months (depending on skeletal sites) . Normal lengths of long bones in the present rats are consistent with other animal experiments [25–27].
In conclusion, enterocystoplasty in non-acidotic rats with normal renal function neither impairs skeletal growth nor bone metabolism, which agrees with most clinical series. When the ileocaecal segment was resected there was significant loss of bone mass. Changes were confined to the metabolically highly active cancellous compartment with rarefaction of the trabecular network, reducing the endosteal bone surface and expanding the marrow cavity area. These features resemble metabolic bone disease, most likely as a consequence of intestinal malabsorption.
While the rat model is well established in preclinical osteoporosis research, it is not standardized in urinary diversion, with numerous physiological limitations. Children undergoing such surgery should be followed prospectively to determine whether similar problems occur in clinical reality.
In memory of Professor Lis Mosekilde; while conducting this work we lost a great colleague and collaborator, and the field of bone research has lost a great scientist. Supported by grants from Deutsche Forschungsgemeinschaft (GE 973/1–1), Bad Godesberg, Germany, and the Incontinence Research Trust, London, England.
dual-energy X-ray absorptiometry
peripheral quantitative CT
bone mineral content
bone mineral density