GYNECOLOGY
Urinary diversion in patients undergoing pelvic exenteration JAMES W. ORR, JR., M.D. HUGH M. SHINGLETON, M.D. KENNETH D. HATCH, M.D. PEYT0;\1 T. TAYLOR, M.D. J. MAX AUSTIN, JR., M.D. EDWARD E. PARTRIDGE, M.D. SENG JAW SOONG, PH.D. Birmingham, Alabama Between October, 1969, and April, 1981, gynecologic oncologists at the University of Alabama Medical Center in Birmingham have performed 119 pelvic exenterations. One hundred fifteen of these patients had a concurrent supravesical urinary diversion. Fifty-six patients (48.7%) had an anterior exenteration and 59 patients (51.3%) had a total exenteration. An ileal segment was used as a conduit in 97 patients while a segment of transverse colon was used in 16 patients. Two patients had sigmoid conduits. Eighty-iive patients (73.9%) had the intestinal anastomosis and conduit constructed with gastrointestinal staplers. Stapler use shortened the mean operating time for the exenterative procedure by approximately 30%. No increase in postoperative gastrointestinal complications was noted. Urinary diversion performed as part of a pelvic exenteration has been associated with short- and long-term complications. The use of ureteral stents and the gastrointestinal staplers shortens the procedure without predisposing the patient to major urologic complications. The use of a segment of unirradiated bowel (transverse colon) in conjunction with these techniques constitutes the preferred method of supravesical urinary diversion in patients undergoing a pelvic exenteration. (AM. J. 0BSTET. GYNECOL. 142:883, 1982.)
From the Division of Gynecologic Oncology, Department of Obs!Ptrics and Gynecology, and Department of Biostatistics, University of Alabama Medical Center in Birmingham. Receivedfor jnlblication September 9, 1981. Revised November 2, 1981. Accepted November 9, 1981. Reprint requests: James W. Orr, Jr., M.D., Department of Obstetrics and G.J•necology, Division of Gynecologic Oncology, University of Alabama Medical Center in Birmingham, University Station, Birmingham, Alabama 35294. )02-9378/82/070883+07$00.70/0 © 1982 The C. V. Mosby Co.
support resulting from modern anesthetic techniques, intraoperative and postoperative monitoring (pulmonary artery catheters, arterial cannulation), fluid and blood component therapy, and antibiotics has decreased postoperative morbidity and mortality associated with exenterative surgery. With this decreased incidence of sepsis and hemorrhage, increasing attention has focused on problems associated with urinary diversion. Recent reports have emphasized the importance of intestinal segment selection,t-:J the use of gastrointestinal staplers, 4 and the IMPROVED PERIOPERATIVE
883
884
.\pril L J9,,:!
Orr et aL .\01
Table I. Indications for pelvic exenteration, October, 1969 to March, 1981 Cervical carrinoma: Recurrent Radiation necrosis Endometrial carcinoma Vaginal carcinoma Vulvar carcinoma Other Total
101
99 ~
5
7 3 3 119
use of nreteral stents:' to minimize the postoperative complications associated with urinary diversion during pelvi(· exteneration. This report details and examines these factors and others in relation to complications associated with urinary diversion performed in conjunction with exenterative surgerv for gynecologic malignancy.
Methods and material Between October, 1969, and March, 1981, gynecologic oncologi~ts at the University of Alabama Medical Center in Birmingham have performed 119 pelvic cxenterations. The major indication for this procedure was recurrent cervical malignancy; however, I 6.8% of these procedures were performed for other indications (Table !). Although the methods of urinary diversion are well described .~--tithe procedures used at this institution dif~ fer in several respects. All ureteral anastomoses were perf
Results One hundred fifteen patients had a urinary conduit performed during a pelvic exenteration. The major-
J
(Jb,tel. (~Vtlt'COJ
ity (9H.3cX) had received prior pelvic radiot ht•tix patienh HR71Jr) had an anterior exenteration. ,md S9 rMtJcnb (!i l.:{c/(.) had a total exenteration. An 1kal ;,cgmenr 1\;t, used in the urinarY diversion of 97 patient' (H4.'\~:; ). while a segment of transverse colon was used itt I (i patients (13.9c,t). Two patients (I. 7~:.;) had sigmoid conduits (Table II). Eighty-five patients (73.9%) had the intestinal ana~~ tomoses and conduit constructed with gastrointestinal staples. Staples were used in 71.1 o/c of the ileal conduib and in all of the transverse colon conduits. :":either ol the sigmoid conduits incorporated staples (fable IIi. The median age (51 vears; range, 27 to 76) and the median weight ( HH pounds; range, 7-l to 268 pounds) were not significantly different between patients with an anterior or total exenteration and patients with an ileal or transverse colon conduit or between patients with or without stapled anastomoses. The mean operating time for an amerior exenteration was significantly shorter than that for a total exenteration tleast square of means, p < 0.001). The usc ol a gastrointestinal stapler shortened the operati\e time regarclles' of the magnitude of the procedure ur the choice of intestinal segment (Table Ill). fhere wa' 1111 increase in gastrointestinal complications related to staple use. The postoperative course in :21 patients (20.9 1;( J W
Volume 142 Number 7
Urinary diversion in pelvic exenteration
885
Table II. Distribution of intestinal segments used in exenterations with or without the gastrointestinal stapler Procedure
Anterior exenteration
Ileal conduit: Staples No staples Transverse colon conduit: Staples No staples Sigmoid colon conduit: Staples No staples Overall(%) Staples No staples
52 36 16
Anterior*
Intestinal segment
4
Ilealt
Total*
+
Ilealt
+
Ilealt Transverse colon
+
2 (1.9%)
2
56 (48.7%)
Transverse colon
16 (100%) 0
0 2
40 16
Ileal t
16 (13.9%)
12
0 0 0
+
69 (7l.l%) 28 (28.9%)
12 0
4 0
Staples
97 (84.3%)
45 33 12
0 2 59 (51.3%)
115 (100%)
45 14
Table III. Operative time related to type of exenteration, type of intestinal segment, and stapler use Type af exenteration
Overall(%)
Total exenteration
Operative time in hours (mean, median, range)
4.9 ± 0.8 5.0 (3.4-7.0) 6.8 ± 1.3 6.7 (5.2-9.25) (5.2 ± 0.7 5.1 (4.6-6.1) 5.4 ± 0.9 5.3 (3.3-7.8) 6.4 ± 1.3 6.6 (4.3-8.3) 5.9 ± 1.3 5.9 (4.4-9.2)
*Anterior exenteration was significantly shorter than total exenteration (P <0.001). tStaples were associated with a significantly shorter operative time (T test: anterior, P 0.0001; total, P = 0.005). course of the patients with early urinary leaks was usually complicated. Two patients developed a pelvic abscess, one patient had a concurrent enterocutaneous fistula, and one patient developed a small bowel obstruction and had an intervening operative procedure prior to developing a urinary fistula. One patient (0.8%) had necrosis of the conduit secondary to vascular insufficiency of the isolated segment of small bowel. The one patient with a late urinary fistula had developed a pelvic abscess secondary to a complete stricture of the pelvic defect. Three patients (2.6%) developed a urinary fistula/
85 (73.9%) 30 (26.1%)
Table IV. Pelvic exenteration complicated by pyelonephritis Postaperative Rehospitalization pyelonephritis for pyelonephritis Type of conduit
No.
Ileal (97t) Transverse colon (16) Sigmoid colon (2) Total patients (115)
21 3
~ % 21.6 18.8
No.
16 0
0 24
I
No.j
%*
16
%
37 38.1 3 18.8
16.5 0
0
0 20.9
Overall incidence
13.9
40 34.7
*Percent of all patients. tNo. of patients. *Sixteen patients with 22 admissions.
Table V. Pelvic exenteration complicated by postoperative pyelonephritis: Organisms Organism Pseudomonas Serratia E. coli Citrobacter Enterobacter Proteus
!
% 35 20 15 10 10 lO
leak that may have been solely related to a failure of closure of the proximal end of the conduit segment. Two of these patients (2.4%) had a conduit closed with staples and one patient (3.2%) had the conduit closed with sutures. The management of patients with urinary fistula varied. Six patients had a second operative procedure. Three of these patients (50%) died. Three patients had spontaneous closure of the urinary leak at 2, 4, and 16 months, respectively (Table VIII). The patients with
886
April l, !9K2
Orr et aL .\IlL
Table VI.
~on-cancer-related
Operation:fistula interval (da_w)
Fistula site
Cornplicating factors Ileal Ileal Ileal
None
HI
'T'otal Anterior Total
I)'"": -I
Defect
Total
.Ilea!
None
~9
Cutaneom Cutaneous
Anterior Total
Ileal Ileal
Enterocutaneous fistula None
Defect Small bowel Defect
Total Anterior Anterior
Ileal Ileal Ileal
De fen
Total
Ileal
Pelvic abscess Conduit necrosis Diabetes, rectovaginal fistula, pelvic abscess Pelvic abscess
30 31 35
60 220
Ohster. t'mewl.
fistulas and urinary leaks
Defect Defect Cutaneous
15 !?>
.l
spontaneous closure of the fistula frequently had an uncomplicated postoperative course (Table VI). None of the cutaneous fistulas closed spontaneously and reoperation on these patients was associated with a mortality rate of 67%. Late problems referable to urinary diversion were present in 24 patients (20.9%) and represented 61.60!· and 2l.OC1r of patients who required rehospitalization or a second operation for nonmalignant indications (Table IX). Sixteen patients (22 admissions) representing 41% ot all patients rehospitalized for a nonmalignant indication were readmitted for pyelonephritis. All of these patients had ileal conduits (Table IV). The causati\e organisms were not different from those in patients with acute postoperative infections. Four patients (:-~.5'7[), all with ileal segments, later underwent a conduit stoma revision. Three patients underwent a second operation for nephrectomy or ureteral stricture. One patient required a second anesthetic for stent remoYal via the conduit (Table IX). Two patients (2.4c,.t.) with staples used to close the conduit have passed urinary stones. ~either has had a deterioration of renal function. Indicated follow-up pyelograms and roentgenographv of the conduit usually revealed mild hydronephrosis and ureteral reflux. A single patient with a transverse colon conduit experienced a metabolic abnormality secondary to urinary diversion. An additional patient with an ileal conduit had an unexplained deterioration of renal function (serum creatinine,> 1.5 mg/ l 00 ml) which developed 8 years after diversion.
Comment Since Bricker's" initial description of urinary diversion bv conduit, numerous publications have appeared describing the technical aspects and short- and long-
None
Intervening operation for small bowel obstruction
Treatment Spontaneous closure at 6 month~ Spontaneous closure at 16 months Reoperation, transverse colon conduit Reoperation. transverse colo11 conduit, postoperative death Reoperation. postoperative death Pulmonary t'mbolus. postoperative death Spontaneous closure at 2 months Reoperation, new ileal conduit Reoperation, nephrostomies, postoperative death Reoperation, conduit repair, percutaneous nephrostomies
term complications associated with this procedure. Many of these papers include diversiom performed in children or adults for nonmalignant indications. Frequently, these operations were performed as the primary procedure in nonirradiated patients. Although almost all of the patients t9K.2Stl in the present report had received radiotherap\ as the initial treatment for a pelvic malignann. the rime/ dose relationship varied, which made it difficult to correlate morbidity and complications wirh pre\·ious radiotherapy. It should be noted that healing of these tissues is altered, predisposing these patients to increased complications. Historically, problem:; associated with urinar} diver· sion hm·e been divided into short- and long-term complications: however, few recent reports havt' a large number of patients with different n pes of conduits.1· ~. ''· 10 · 11 The purpose of this repon was to evaluate these short- and long-term complications in a homogeneous group of patients and to relate the effect of use of intestinal segment (irradiated \·ersus nonirradiated) and gastrointestinal stapler to these cmnplications. Postoperatively, acute p)'elonephritis was a common problem in this series. Although the introduction of a foreign body (ureteral stent) into the collecting system might theoretically increase the risk of urinary infection, the incidence of this complication was not different from incidences in reports where stents were nor used. 7 The most serious complication of urinary di\·ersion with exenteration is the development of a urinary leak or fistula. The reported incidence of this complication varies between 5% and 22%.t. 2 • :;. !1- 12 depending on the type of intestinal segment, 1a the magnitude of the procedure,1z and the use of ureteral stents.:' Although subclinical leaks are not unusual and often dose ~pon-
Volume 142 :'>lumber 7
Urinary diversion in pelvic exenteration
Table VII. Pelvic exenteration: Non-cancer-related urinary fistulas or leaks No. of patients Age (yr): Mean Median Range Weight (pounds): Mean Median Range Postoperative interval (days): Mean Median Range
10 (8.7%) 52.2 52.5 38-65
Table VIII. Pelvic exenteration: Management* of urinary fistulas Operative management (50% mortality) Spontaneous closure (2, 4, and 16 months)
~fransverse
colon
6 Patients 3 Patients
*One patient died of other complications of exenteration. 128.4 116 91-188 29.1 29 15-60
Table IX. Pelvic exenteration: Rehospitalization secondary to urinary complications
Fistula site:
Defect Cutaneous Small bowel Type of exenteration: Anterior Total Intestinal segment: Ileal
887
6 (5.2%) 3 (2.5%) 1 (0.9%) 4 (7.1%) 6 (10.8%) 10 (10.3%) 0 (0%)
taneously, 14 the problems of a persistent urinary leak are con1plex. Additionally, some reports·5 combine urinary leaks and fistulas into a single group, implying that the former may close spontaneously. If patients with spontaneous closure are omitted, only seven patients (6.1 %) in this series developed non-malignancy-related urinary fistula. Regardless of the definition, it would be helpful to identify patients at high risk to develop a urinary fistula. It appears that patients with previous pelvic radiotherapy or a malignancy are at increased risk. 11 - 14 Since the majority of patients who undergo a pelvic exenteration for a recurrent gynecologic malignancy fall into both of these categories, it would be helpful to define a high-risk subgroup. A common association with a urinary leak in this series was the development of a pelvic abscess. All three patients in this series with a pelvic abscess developed a urinary fistula. The information from this series differs from that of one other report 10 as serious morbidity was not increased in older or heavier women. In fact, the median weight of patients who developed urinary fistula in this series was less than that of the remaining patients. This weight difference was not explained by a chronic state of malnutrition. The fact that the urinary fistula rate was not significantly different in patients with an anterior (7.1 %) or total (10.2%) exenteration indicates that a more extensive procedure did not place the patient at a higher risk. The most striking factor in this series is that all the fistulas occurred in patients with an irradiated ileal
Patient)
Etiology
Pyelonephritist Conduit stomal revision Nephrectomy Ureteral stricture Stem removal Metabolic abnormality
No.I
%*
16 4
13.9 3.5
2 1
Patients rehospitalized (non-cancerrelated indicatiom) (%)
Patients reoperated upon (non -cancerrelated indications) ( %)
41
10.3
10.5
0.9
5.1 2.6
5.3 2.6
0.9
2.6
2.6
0.9
2.6
1.7
*Percent of all patients. tSixteen patients with 22 admissions. segment. The distal ileum, in contrast to the transverse colon, is usually immobile and during pelvic radiotherapy is likely to receive a high radiation dosage which may subsequently affect healing. The high fistula rate in ileal conduits (10.3%) was clinically apparent to the oncologists at this institution and a general trend to the routine use of colon conduits was established prior to this report. Ureteral stents have been routinely used in all patients in this series undergoing urinary diversion with pelvic exenteration. The ease of ureteral anastomosis Over a stent and protection of an anastomotic site c!inically outweighs the apparent risk of renal parenchymal damage or stem obstruction. The ureteral stems are usually spontaneously expelled after the anchoring suture is absorbed between the fifteenth and twentieth postoperative days. Because stents are routinely used, it is impossible to compare these benefits and effectiveness among patients in this series. However, patients in the present series with a nonirradiated intestinal segment (transverse colon) and ureteral stems had a decidedly lower fistula rate (0) when compared to patients in other series (9.25%) where ureteral stems were not used.!. 3 The absence of acute anastomotic stricture in this series compares favorably with that of previous reports.t. 7 Our enthusiasm for the use of ureteral stems is shared by at least two other groups. 2 • 10 Although intravenous pyelograms and x-ray films of
888
.\p1II l
Orr et al. '\111.
the conduit may be helpful, the exact point of a urinary leak is often difficult to localize. Other than the ureteral anastomosis the proximal end of the conduit is at risk. Eighty-five patients (78.2o/c) had this portion of the conduit closed with a gastrointestinal stapler. Although some authors1.; do not advocate its use, we are in agreement with others 4 that the benefits of staple use outweigh the risks in exenterative surgen. In this series the incidence of urinary leaks related to possible conduit failure without concurrent complications was not different between sutured (3.2';k) and stapled (2.3'7c) closures. The gastrointestinal stapler allows one to minimize bowel trauma and decrease peritoneal contamination while using a monofilament, minimally reactive permanent material that maximizes r·evascularization in the intestinal anastomoses. The additional benefit of decreasing operative time by approximately 30'7c in a long procedure should diminish fatigue and decrease the likelihood oftechnical errors later in the procedure. The management of a urinarv leak is criticaL Frequenth, the patients in this series had other concurrent serious problems such as a pelvic abscess or small bowel obstruction requiring intensi\·e postoperative support. The 'ur2·ical mortalitv of a reoner:cttion in thi~ series (5or;:) was sig11iticant and is similar to that of other reports .1"· 11 Conduit-cmaneous fistulas were ominous (67cX mortality). Since 30'7c of the fistulas closed spontaneously and reoperation carried a significant mortality, current management at this institution includes careful observation and maximal medical support. Reoperauon is usually resen·ed for those patients with complex fistulas, recurrem infection, or life-threatening deterioration of renal function. Late non-cancer-related problems referable to the urinarY tract constitute substantial morbidity in paticnl:'. undergoing a pelvic exenteration. In this series 61 rx of the patients who required rehospitalization for nonmalignant indications had problems associated with the urinary tract. Late pvelonephritis was the most common reason for rehospitalization. Interestingly, while the experience is small. no patient with a trans,·er~e colon conduit has required rehospitalization for acute pyelonephritis. This is in disagreement with resuits of others 7 who reported a i 7 .4'/c incidence of late pyelonephritis in patients with colon conduits. The higher incidence of urinary infection in ileal conduits may be related to the fact that the urine pH tends to be less acidic or even alkaline .16 Since colonization of the <.~
--
--
I
--
-
-
J
~-
----
---
-----
--
--
J. Uhstt'l.
1%~
Cvnecoi
conduit is common,"' w the loss of a IH>nntnn·'· it is often a late complication of urinarv diversion anclma\ be related to chronic infection or prior radiation. -rhe use of a large-caliber lumen (colon conduit) rna\ protect against this complication. Three additional patients (2 .5%) required operative procedures for late ureteral strictures. All of these patients had ileal segments used in the urinan diversion. There is some controversy concerning the long-term complication of' urinarY calculi associated with stapled conduitclosuresY· 11 '· 1' The stone' tend to be small and to pass spontaneously and apparent]\ are ol little importance unless a subsequent conduit-\'esical ;mastomosis is planned. ~either of the patient-. who dneloped conduit stones in this series has had a subsequent deterioration of renal function or tn_·1n-rem 'Ymptomatic infection. In conclusion, the supra \·esical u rinan di\ t-rsion perf(>rmed with a pelvic exenteration is fraught with short.- and long-term complications. l.'reteral stents and the gastrointestinal stapler benefit both the surgeon and the patient. The use of unirradiated bowel (trans\-erse colon) as the intestinal segment carnes less risk of anastomotic failure, late stenosis, or infection without adding technical difhculty and should decre
I
_____ I _ · . ·
'
_
I
'T'L
'
_
'_1 __
REFERENCES
I. Schmidt, J. D., Buchsbaum, H. J., and Jacobo, E. C.: Transverse colon conduit for supravesical urinary tract diYersion, Urology 8:542, 1976.
2. Schlesinger. R. E., Bailon, S. C .. Wau·ing. W. (;., and Moore. J. G.: The choice of an intestinal segment for a urinary conduit, Surg. Gynecol. Obstet. 148:45. 1979.
~
Volume H2 Number 7
3. Schmidt, J. D., Hawtrey, C. E., and Buchsbaum, H. J.: Transverse colon conduit: A preferred method of urinary diversion for radiation-treated pelvic malignancies, J. Urol. 113:208, 1975. 4. Delgado, G.: Cse of the automatic stapler in urinary conduit diversions and pelvic exenterations, Gynecol. Oneal. 10:93, 1980. 5. Rutledge, F. N., Smith, J. P., Wharton, J. T., and O'Quinn, A. G.: Pelvic exenteration: Analysis of 296 patients, AM . .J. 0BSTET. GYNECOL. 129:881, 1977. 6. Bricker, E. M.: Bladder substitution after pelvic evisceration, Surg. Clin. North Am. 30:1511, 1950. 7. Morales, P., and Golimbu, M.: Colonic urinary diversion: 10 years of experience, J. Urol. 113:302, 1975. 8. l'\elson, J. H.: Atlas of Radical Pelvic Surgery, New York, 1968, Appleton-Century-Crofts, pp. 181-191. 9. Symmonds, R. E., Pratt,}. H., and Welch,J. S.: Exenterative operations: Experience with 118 patients, AM. ]. 0BSTET. GYSECOL. 101:66, 1968. 10. Swan, R. W .. and Rutledge, F. N .: Urinary conduit in pelvic cancer patients: A report of 16 years' experience, AM. J. 0BSTET. GYNECOL. 119:6, 1974. II. Barber, H. R. K., and Brunchwig, A.: Urinary tract fistulas following pelvic exenterations, Obstet. Gynecol. 28:754, 1966.
Urinary diversion in pelvic exenteration
889
12. Schoenberg, W., and Mikuta, J. J.: Technique for preventing urinary fistulas following pelvic exenteration and ureteroileostomy,]. Urol. 110:294, 1973. 13. Fallon, B., Leaning, S., Hawtrey, C. E., Lifshitz, S. G., and Buchsbaum, H. J .: Urologic complications of pelvic exenteration for gynecologic malignancy, J. Urol. 122:158, 1979. 14. Bagley, D., Glazier, W., Osias, M., and Lytton, B.: Retroperitoneal drainage of ureterointestinal conduits, j. U roi. 121:271, 1979. 15. Bergman, S.M., Sears, H. F., andJavadpour,l\'.: Complication with mechanical stapling device in creation of ileoconduit, Urology 7:71, 1978. 16. Guinan, P. D., Moore, R. H., Neter, E., and Murphy, G. P.: The bacteriology of ileal conduit urine in man, Surg. (;ynecol. Obstet. 134:78, 1972. 17. Bricker, E. M.: Current status of urinary diversion, Cancer 45:2936, 1980. 18. Karamcheti, F., O'Donnell, W. F., Hakala, T. R., Schwentker, F. N ., and Steichen, F. M.: Autosuture ileal conduit construction: Experience in 110 cases, J. Urol. 120:545, 1978.