In utero surgical treatment of fetal obstructive uropathy: A new comprehensive approach to identify appropriate candidates for vesicoamniotic shunt therapy

In utero surgical treatment of fetal obstructive uropathy: A new comprehensive approach to identify appropriate candidates for vesicoamniotic shunt therapy Mark Paul Johnson, MD,"' b. Timothy P. Bukowski, MD,d Claude Reitieman, MD,d Nelson B. Isada, MD," Peter G. Pryde, MD," and Mark I. Evans, MD·' b. c C

Detroit, Michigan OBJECTIVES: We present a new approach to prenatal evaluation and prediction of renal function and long-term outcomes for fetuses considered candidates for in utero surgery. STUDY DESIGN: A review of 34 cases of obstructuve uropathies was made, with analysis of our approach to prenatal evaluation, the predictive value of urinary electrolytes and protein analysis in identifying renal dysplasia, and the outcomes of cases with and without in utero surgical intervention. RESULTS: Ten of 11 pregnancies that were terminated had dysplasia; one female had cloacal dysgenesis. In three of eight unshunted cases megacystis resolved after vesicocentesis. Eight of 15 shunted fetuses survived. Of those predicted to have good postnatal outcome, six of six demonstrated good postnatal renal function. Of those predicted to have poor outcomes, two of two had poor postnatal renal function and dysplasia. Seven of 15 shunted fetuses died, four of four predicted to have dysplasia were confirmed, and three died of nonrenal causes. CONCLUSIONS: Our prenatal approach successfully differentiated fetuses with underlying renal dysplasia from those without irreversible damage who subsequently benefited from in utero vesicoamniotic shunt placement. (AM J OSSTET GYNECOL 1994;170:1770-9.)

Key words: Obstructive uropathy, prenatal diagnosis, fetal urine, renal dysplasia, fetal therapy Fetal obstructive uropathies involve a heterogeneous group of developmental abnormalities that result in partial or complete obstruction of urinary outflow at any level of the urinary tract. The lack of amniotic fluid leads to pulmonary hypoplasia, characteristic Potter sequence appearance, and an extremely high mortality rate. Two major groups have been identified: (I) those with normal to mildly diminished amniotic fluid volume (partial obstruction) and (2) those with oligohydroamnios or anhydramnios (complete obstruction). Experience supports the conclusion that active intervention for group 1 is not warranted because the outcome will likely be good with only neonatal intervention required. The prognosis for fetuses with complete obstruction is quite different. The natural course of events involves

From the Center for Fetal Diagnosis and Therapy, Departments of Obstetrics and Gynecology," Molecular Biology and Genetics," and Pathology,' Hutzel Hospital, and the Department of Pediatric Urology,d Childrens Hospital of Michigan, Wayne State University School of Medicine. Presented at the Sixty-first Annual Meeting of The Central Association of Obstetricians and Gynecologists, White Sulphur Springs, West Virginia, October 28-30, 1993. Reprint requests: Mark Paul Johnson, MD, Division of Reproductive Genetics, Center for Fetal Diagnosis and Therapy, Department Obstetrics and Gynecology, Hutzel Hospital/Wayne State University, 4707 St. Antoine Blvd., Detroit, M148201. Copyright © 1994 by Mosby-Year Book, Inc. 0002-9378/94 $3.00 + 0 6/6/55034

1770

progressive oligohydramnios, secondary fetal deformations to the face and extremities, pulmonary hypoplasia, and renal dysplasia. Pioneering work by the Fetal Therapy Group at the University of California, San Francisco, with a fetal sheep model clearly showed that early, prolonged obstruction leads to hydronephrosis and eventual renal fibrosis and cystic dysplasia. 1 The degree of irreversible renal damage was directly correlated with age of onset and duration of obstruction. In addition, if early obstruction was relieved after a short to moderate duration, pulmonary hypoplasia and renal dysplasia could be prevented. These experiments led to the development of the vesicoamniotic shunt, in which a catheter is placed in utero under ultrasonographic guidance into the fetal bladder while the distal end is positioned in the amniotic space. Such shunting bypasses the obstruction, allowing fetal urine to flow into the amniotic space and restoring fluid volume and preventing secondary damage to the fetus. Over the past decade a great deal has been learned about fetuses with these anomalies. However, the ability to predict which fetuses will actually benefit from in utero surgery has been limited. Our center has been very active in this area of interventive fetal therapy, and we have developed a systematic approach to the evaluation, treatment, and follow-up of such patients. This study is the first presentation of our new approach to

Volume 170. Number 6 Am J Obstet Gynecol

the prenatal evaluation and prediction of underlying renal function and long-term outcomes for fetuses considered possible candidates for interventive therapy. Material and methods

Thirty-four women from institutions across the United States whose pregnancies were diagnosed as complicated by megacystis and hydronephrosis were referred for prenatal evaluation. Analysis consisted of retrospective review in six cases and prospective comprehensive analysis in the remaining 28. The only difference between these groups was that the retrospt ctive cases underwent a single urine sampling by vesicocentesis, whereas the prospective group underwent se· rial samplings. Gestational age ranged from 14 to 24 weeks on the basis of last menstrual period, confirmed by ultrasonographic measurements. Only cases considered to have true outlet obstruction described ultrasonographically as megacystis, bilateral hydronephrosis, and decreased amniotic fluid volume were included in the study. In cases < 16 weeks, a moderate decrea~e in amniotic fluid volume was accepted, whereas > 16 weeks required oligohydramnios for inclusion. Six cases were initially evaluated at 14 to 16 weeks, 10 at 17 to 18 weeks, six at 19 to 20 weeks, three at 21 to 22 weeks, and nine at 23 to 24 weeks. No cases > 24 weeks at initial visit were included. All cases of upper urinary tract obstruction (uteropelvic or uterovesical junction) were excluded. All families were counseled extensively about the various causes, the expected clinical course, and the associated prognosis for such obstructions. Our approach to these fetuses includes (1) detailed ultrasonographic evaluation, (2) rapid fetal karyotype analysis, and (3) serial evaluations of electrolytes and protein in the fetal urine. In cases of oligohydramnios or anhydramnios, an amnioinfusion was routinely performed with warmed 5% dextrose in lactated Ringer's solution with volumes of 150 to 250 ml, depending on gestational age. The infusion was performed to assist visualization, restore amniotic fluid volume to low-normal levels, and to prevent secondary deformations of the face and limbs. The fetus was screened for other congenital anomalies, and the urinary tract was evaluated in great detail. In one case a fetus was found to have an isolated mild kyphoscoliosis of the upper thoracic and cervical spine. Pediatric neurologists considered this to be a minor abnormality that should not exclude the fetus from evaluation protocols. All other fetuses found to have other ultrasonographically identifiable congenital malformations on initial ultrasonography were excluded. The urethra was evaluated for origin of obstruction (anterior urethral valves for megaurethra, atresia or posterior urethral values for vesicourethral junction dilation or "keyhole sign") and for presence of male

Johnson et al.

1771

genitalia. The bladder was examined for degree of dilation, presence of urachal abnormalities, degree of thickening in bladder neck and dome after vesicocentesis, and for evidence of gross reflux at the ureterovesical angles. Ureters were evaluated for bilateral involvement, presence and degree of dilation and distortion, and extent of pyelectasis. Kidneys were evaluated for degree of hydronephrosis, size for gestational age, echodensity, thickness of cortical parenchyma, and presence of discrete cortical cysts. After discussion of findings and their correlation with prognosis, fetal karyotyping was undertaken in the majority of cases by transabdominal chorionic villus sampling and in a few cases by amniocentesis before amnioinfusion. One fetus was found to have female (46,XX) chromosomal complement. The family was counseled as to likely associated cloacal malformations, and they elected to terminate the pregnancy. The fetus was confirmed to have complex cloacal anomalies. We are not aware of any such cases that have benefited from vesicoamniotic, or in this case, cloacal-amniotic shunting, and therefore only male infants are considered as candidates for such therapy. The presence of irreversible fetal renal damage was evaluated by studying fetal urine electrolytes and proteins and rate of bladder refilling after vesicocentesis. Fetal urine was evaluated for sodium, calcium, osmolality, ~2-microglobulin, and total protein levels. The rate of bladder refilling in the 30 to 60 minutes after vesicocentesis was subjectively recorded as an indicator of patency of upper tracts and possible urine production. With the exception of the retrospective cases, urine values were obtained and compared from a minimum of three vesicocenteses performed at 48- to 72hour intervals, and a documented sequential fall in values was considered as evidence of potential improvement with chronic bladder drainage as offered by vesicoamniotic shunting (Fig. 1). Sequential rise in values was felt to reflect progressive dysplasia, which would not benefit from long-term drainage and therefore shunt placement. Electrolyte and protein cutoff values used in considering a fetus for shunting were based in part on our experience and on values reported in the literature as being associated with good long-term neonatal prognosis (Table I). Values from the last fetal urine sample and documentation of progressive improvement in values with sequential bladder drainage were used in deciding which fetuses represented good candidates for shunting. Urine values with sodium < 100 mg/dl, calcium < 8 mg/dl, osmolality < 200 mOsm/L, ~2-micro­ globulin < 4 mg/L, and total protein < 20 mg/dl were thought to represent the absence of irreversible renal dysplasia, and these fetuses were thought to represent the optimal candidates for vesicoamniotic shunts. Fetuses with urine values above these cutoff values were

1772 Johnson et al.

. June 1994 Am J Obstet Gynecol

250

A 200 150 100 50. 0

Ca

Na

T.P.

Osm

B2m

Na

T.P.

Osm

B2m

250 B 200 150 100 50 0

Ca

Fig. 1. Sequential fetal urinary electrolyte and protein values. In A the fetus demonstrates sequential improvement in all parameters; this pattern is associated with absence of underlying renal dysplasia and good predicted postnatal outcome after shunt placement. In B the fetus demonstrates sequential worsening in parameters associated with significant irreversible underlying renal fibrosis and dysplasia and poor postnatal outcome. Ca, Calcium; Na, sodium; T.P., total protein; Osm, osmolality; B2m, 132-microglobulin.

Table I. Fetal urine electrolyte and protein cutoff values used in evaluating underlying renal function Sodium (mg/dl) Calcium (mg/dl) Osmolality (mOsm/L) 132-Microglobulin (mg/L) Total protein (mg/dl)

< 100 <8 <200 <4 <40

thought to have irreversible renal damage; the families were counseled that shunting would not likely improve the long-term prognosis and in general were not encouraged to undertake invasive surgery. Vesicoamniotic shunting was performed within 2 weeks of the initial visit by means of Rodeck double pig-tailed shunt (Rocket of London, Watford, U.K.). All

shunts were placed under continuous ultrasonographic guidance, near the midline between the fetal pubis and umbilical cord insertion, with identification of the course of the umbilical vessels lateral to the bladder with color Doppler flow imaging just before the fetal bladder was entered. All shunts were successfully placed and documented to be functioning before the patient returned home. Patients were reevaluated ultrasonographically weekly for 4 weeks and then every 2 weeks thereafter for shunt placement, function, and amniotic fluid volume. Whenever possible, before shunt placement patients were referred to members of the fetal therapy group in pediatric urology for antenatal consultation as to neonatal evaluation and treatment modalities and longterm prognosis and their experience with fetuses who did and did not undergo prenatal shunting.

Johnson et al.

Volume 170, Number 6 Am J Obstet Gynecol

Table II. Summary of cases with obstructive uropathy evaluated Total No. of fetuses evaluated No. of pregnancies voluntarily terminated No. of fetuses observed Obstruction resolved In utero or neonatal death No. of fetuses with vesicoamniotic shunts

34

11 8 3

5

15

Six shunts were dislodged into the amniotic space by the fetus and were replaced. Three shunts were displaced intraperitoneally, resulting in urinary ascites. This was successfully corrected in all cases by placement of a peritoneal-amniotic shunt. Postnatal treatment and follow-up consisted of vesicostomy (minimum 2 years) for posterior urethral valves or urethral atresias and pediatric urology clinic visits every 3 months for 1 year and then every 4 to 6 months thereafter. Good outcome is defined as a survivor with a serum creatinine level at 1 year of S 1.0. Poor outcome was defined as a perinatal death with necropsyproved renal dysplasia, or a survivor with post-I-year creatinine level > 1.0. The youngest survivor is 18 months old and the oldest 5.5 years old. If the patient terminated the pregnancy or the infant died, a detailed study of the entire urinary tract was performed on necropsy. Results

Thirty-four fetuses with true outlet obstruction were evaluated (Table II). Eleven pregnancies were terminated. One fetus was found to be female (46,XX) with cloacal dysgenesis, and termination of the pregnancy was chosen. Major cloacal malformations were confirmed at necropsy. The remaining 10 pregnancies were terminated on the basis of prenatal ultrasonographic and renal function evaluation. Advanced renal fibrosis' and dysplasia were confirmed at necropsy in all cases. One fetus with Klinefelter syndrome (47 ,XXV) for whom the parents chose to continue the pregnancy met criteria and underwent vesicoamniotic shunting. He is now a healthy I8-month-old infant whose creatinine level is 0.6. No other chromosomally abnormal fetuses were included in this study. Eight fetuses were followed up without attempted shunting. In three of these cases the megacystis failed to recur and the obstruction apparently resolved after initial vesicocentesis that completely drained the fetal bladder. In all three cases resolution occurred between 15 and 18 weeks. Five fetuses demonstrated poor ultrasonographic and renal function values, but the parents elected to continue the pregnancy in spite of the poor prognosis. Four of these resulted in intrauterine fetal death before 30 weeks' gestation, and one died < 72 hours after birth of respiratory insufficiency caused by

1773

Table III. Outcome in fetuses who underwent vesicoamniotic shunt placement: fetal or neonatal losses Total No. of fetal or neonatal losses Procedure-related complications Fetuses with predicted good prognosis Fetuses with predicted poor prognosis

7 0 3* 4t

*One case of megacystis and microcolon syndrome, one case of urethral atresia with death at 28 weeks of chorioamnionitis, sepsis, intracranial hemorrage, and hyaline membrane disease, and one case of megaurethra and anterior urethral valves with death at 24 weeks after preterm delivery of twins; none of these cases demonstrated evidence of renal dysplasia on necropsy. tTwo intrauterine fetal deaths, one neonatal death, and one voluntary termination of pregnancy because of progressive oligohydramnios; all cases had documented renal dysplasia on necropsy.

Table IV. Outcome in fetuses who underwent vesicoamniotic shunt placement: long-term survivors Total long-term survivors Good postnatal renal function Predicted good postnatal function Predicted poor postnatal function Poor postnatal renal function Predicted good postnatal function Predicted poor postnatal function

8 6 6

o 2

o 2

pulmonary hypoplasia. This latter case did not undergo shunting until 23 weeks' gestation. In all cases the antepartum diagnosis of underlying renal dysplasia was confirmed on necropsy. Fifteen fetuses underwent vesicoamniotic shunt placement. There were no procedure-related complications that resulted in fetal loss. Seven fetuses died > 1 week after successful shunt placement (Table III). Four of these were predicted by antenatal evaluation to have a very poor prognosis. Two of the four resulted in intrauterine deaths, one infant died of pulmonary insufficiency caused by hypoplasia 48 hours after birth, and in one the pregnancy was later terminated after progressive oligohydramnios caused by apparent lack of urine production. Again renal dysplasia was documented on necropsy in all cases. Three deaths occurred in fetuses who were predicted by antenatal evaluation to have a good renal prognosis. One infant with megacystis-microcolon syndrome died at 5 months of age, and one with urethral atresia died at 28 weeks' gestation (6 weeks after shunt) of chorioamnionitis, sepsis, a large intracranial hemorrhage, and hyaline membrane disease. In the third case, with megaurethra-anterior urethral valve syndrome, the infant died at 24 weeks (3 weeks after shunt) after preterm delivery of this twin pregnancy. No evidence of renal dysplasia was found in any of these three cases.

1774 Johnson et al.

June 1994 Am J Obstet Gynecol

Table V. Ability of biochemical threshold values to predict absence of renal dysplasia Urinary component

Sensitivity

SPecificity

Positive predictive value

Negative predictive value

Sodium < 100 mgldl Calcium < 8 mgldl Osmolality < 200 mOsm/L f32-Microglobulin <4 mg/L Total protein < 20 mgldl

0.56 1.00 0.83 0.17 0.67

0.64 0.27 0.82 0.36 0.91

0.56 0.43 0.71 1.00 0.80

0.88 1.00 0.90 0.44 0.83

There are eight vesicoamniotic shunt survivors (Table IV). Two of these infants were predicted prenatally to have poor renal function "after delivery, and both are presently on dialysis awaiting renal transplant with serum creatinine levels between 1.8 and 2.5. Six infants were predicted to have good renal function postnatally, and all have serum creatinine levels < 1.0 (youngest infant 18 months, oldest 5.5 years). Of the 15 fetuses who underwent vesicoamniotic shunting, two were later found to have cloacal anomalies, three had posterior urethral valves, one had anterior urethral valves and megaurethra, one had megacystis-microcolon syndrome, three had urethral atresias, and five were diagnosed as having prune-belly syndrome.

Comment This study demonstrates that urinary electrolytes and protein factors can successfully identify fetuses with underlying renal dysplasia. In our experience (Table V) urine values obtained on the third or fourth sampling, in which sodium was < 100 mg/dl, osmolality < 200 mOsm/L, and total protein < 20 mg/dl, were all associated with good postnatal renal function and absence of renal dysplasia after successful vesicoamniotic shunt placement. Our study further emphasizes the need for serial vesicocenteses and the demonstration of sequential improvement in urine values before shunt placement (for those with good prognosis). Careful antenatal evaluation of the fetus with true bladder outlet obstruction and decreased amniotic fluid volume can reliably identify fetuses who would benefit from long-term in utero bladder drainage with vesicoamniotic shunting and successfully save these fetuses from an otherwise dismal prognosis and clinical course. Careful attention to excluding fetuses from consideration for shunting who have associated congenital anomalies, female or aneuploid karyotypes, echodense small-for-gestational-age kidneys or renal cortical cysts, and those who fail to demonstrate sequential improvement in urine biochemical values to below the cutoffs outlined in Table I will undoubtedly improve our ability to appropriately select those fetuses who will truly benefit from this form of in utero interventive fetal therapy. Many attempts have been made to determine which

factor or collection of factors within the fetal urine is most reflective of underlying damage and poor longterm prognosis. The early use of sodium, chloride, and osmolality met with some limited success. 2 • 3 During normal renal functioning sodium and chloride are reabsorbed at the proximal tubule, and osmolality falls with reabsorption of electrolytes and proteins. With progressive renal damage proximal tubular function is impaired and values for all three rise. Early reports on outcomes in shunted fetuses on the basis of these predictive factors met with varied success. 4 •6 In all these studies decisions to intervene were based on a single vesicocentesis urine sample. Recently we have shown that values obtained on the initial urine specimen are not predictive of long-term prognosis.' We demonstrated that serial vesicocenteses and use of the third or fourth urine values correlated better with long-term outcome. In addition, we found that a pattern of sequential improvement with each bladder drainage, in addition to absolute electrolyte values, is an important predictor of which fetus will benefit from chronic in utero drainage by shunting. With such an approach we found that sodium and osmolality were good predictors of underlying irreversible renal damage. Nicolini et al. B showed that urinary sodium and osmolality values dropped I day after bladder emptying by vesicocentesis and that serial evaluations were helpful in preventing a false-positive diagnosis of renal dysplasia. In the same study they demonstrated that intraluminal bladder pressures were elevated in fetuses with true outlet obstruction and suggested that tubular reabsorption may be impeded by elevated standing pressure within the urinary tract and that chronic relief of the obstruction and pressure may improve fetal renal function. Nicolini et al. 9 also showed that the concentration of creatinine rises while the values for sodium and osmolality fall in fetal urine with maturation of tubular function in the midtrimester and that appropriate nomograms must be established for comparison when urinary electrolytes are evaluated at differing gestational ages. Additional urine electrolyte parameters that may assist in the antenatal identification of underlying renal dysplasia have been identified. Nicolaides et al.1O investigated the use of sodium, total calcium, urea, and

Johnson et al.

Volume 170, Number 6 Am J Obstet Gynecol

creatinine in fetal urine and showed that values were increased for sodium and total calcium and decreased for urea and creatinine in urine obtained from fetuses with underlying dysplasia compared with that from fetuses without underlying dysplasia. They found either high calcium or high sodium to have the greatest predictive value (positive predictive value 91.3%, negative predictive value 77.7%) for underlying renal dysplasia. Evaluation of fetal proteinuria as a predictor of abnormal renal function has recently gained attention. Mandelbrot et a1. II analyzed many different biochemical factors and found that a sodium value > 70 mEq/L was the most specific predictor of poor fetal outcome, whereas the protein 132-microglobulin when < 2 mg/L was the most sensitive factor in predicting normal postnatal renal function. 132-Microglobulin is almost entirely filtered by the normal glomerulus, similar to creatinine, but unlike creatinine is almost completely reabsorbed by a mechanism of endocytosis and catabolized by proximal tubular cells. Disturbed renal function at the level of the proximal tubules would result in elevated concentrations in the fetal urine. Holzgreve et a1. 12 also compared fetal urine from cases with and without renal dysplasia and showed that elevated levels of total protein positively correlated with the degree of underlying renal damage present. Our group has refined this approach and found that the albumin component of the total protein is the factor that rises most significantly in relation to underlying renal damage and may be a highly sensitive and specific predictor of postnatal renal function. 13 The overall purpose for the detailed prenatal evaluation of fetuses with obstructive uropathies is the identification of those fetuses who will benefit from in utero surgery, while avoiding such invasive intervention in those cases for whom such therapy will offer little to no improvement for outcome. We have periodically reviewed our experience with each component of our diagnostic protocol. No single component of this or any other evaluation provides adequate predictive information for reliably differentiating such fetuses. As noted earlier, much has been written about the use of urine electrolytes and ultrasonographic findings as prognostic factors in predicting long-term outcome. We believe our study emphasizes the importance and power of an integrated approach combining the findings of ultrasonographic features, rapid karyotyping, and urine composition analysis in the identification and selection of fetuses who have not yet progressed to a stage of irreversible renal damage. By combining the data from Tables III and IV our ability to reliably identifY underlying renal dysplasia can be critically assessed. In the 15 fetuses who underwent vesicoamniotic shunting we achieved a 100% success

1775

rate in predicting the presence of significant underlying renal pathologic processes. In the nine fetuses predicted not to have underlying renal dysplasia, six are postnatal survivors with creatinine levels < 1.0 at ~ 1.5 years old. Three of nine fetuses were lost to nonrenal causes, but necropsy demonstrated absence of renal fibrosis or dysplasia. In the six fetuses predicted to have major underlying damage significant dysplasia was confirmed on necropsy or postnatal renal studies. No other report has achieved or reported this kind of success or long-term follow-up information for fetuses who have undergone in utero surgery.' We acknowledge the limitation of our study in that our inclusion and exclusion criteria changed over the course of the study. We feel this reflects the evolution of our knowledge and understanding and the advancement of techniques that improved and expanded our diagnostic capabilities in these cases, further refining our abilities to differentiate fetuses who would benefit from intervention from those in which advanced irreversible damage had made such efforts essentially futile. The development of fetal therapeutic techniques will allow an increasing number of couples now faced with only the options of pregnancy termination or delivery of a child who will likely die from severe renal impairment with the option of directly treating the underlying abnormality before irreversible damage has occurred. Prenatal evaluation of the fetus and the development of good predictive tests and sophisticated surgical procedures are essential to making fetal therapy a reality for future generations of affiicted fetuses. REFERENCES 1. Harrison MR, Filly RA. The fetus with obstructive uropathy: pathophysiology, natural history, selection, and treatment. In: Harrison MR, Golbus MS, Filly RA, eds. The unborn patient. 2nd ed. Philadelphia: WB Saunders, 1990:329-44. 2. Crombleholme TM, Harrison MR, Golbus MS, et al. Fetal intervention in obstructive uropathy: prognostic indicators and efficacy of intervention. AM J OBSTET GYNECOL 1990; 162: 1239-44. 3. Grannum PA, Ghidini A, Scioscia A, CopelJA, Romero R, Hobbins JC. Assessment of fetal renal reserve in low level obstructive uropathy. Lancet 1989;1:281-2. 4. Wilkins lA, Chitkara U, Lynch L, Goldberg JD, Mehalek KE, Berkowitz RL. The nonpredictive value of fetal urinary electrolytes: preliminary report of outcomes and correlations with pathologic diagnosis. AM J OBSTET GYNECOL 1987;157:694-8. 5. Elder JS, O'Grady JP, Ashmead G, Duchett JW, Philipson E. Evaluation of fetal renal function: unreliability of fetal urinary electrolytes. J Urol 1990;144:574-8. 6. Drugan A, Zador IE, Bhatia RK, Sacks AJ, Evans MI. First trimester diagnosis and early treatment of obstructive uropathy. Acta Obstet Gynecol Scand 1989;68:645-9. 7. Evans MI, Sacks AJ, Johnson MP, Robichaux AGIII, May M, Moghissi KS. Sequential invasive assessment of fetal renal function and intrauterine treatment of fetal obstructive uropathies. Obstet Gynecol 1991;77:545-50. 8. Nicolini U, Tannirandorn Y, Vaughan J, Fisk NM, Nicolaidis P, Rodeck CH. Further predictors of renal dysplasia

1776 Johnson et al.

in fetal obstructive uropathy: bladder pressure and biochemistry of "fresh" urine. Prenat Diagn 1991;11:159-6. 9. Nicolini U, Fisk NM, Rodeck CH, Beacham J. Fetal urine biochemistry: an index of renal maturation and dysfunction. Br] Obstet Gynaecol 1992;99:46-50. 10. Nicolaides KH, Cheng HH, Snijders R]M, Moriz CF. Fetal urine biochemistry in the assessment of obstructive uropathy. AM] OBSTET GYNECOL 1992;166:932-7. 11. Mandelbrot L, Dumez Y, Muller F, Dommergues M. Prenatal prediction of renal function in fetal obstructive uropathies.] Perinat Med 1991;19:283-7. 12. Holzgreve W, Lison A, Bulla M, Evans M. Protein analysis to determine fetal kidney function [Abstract]. AM] OBSTET GYNECOL 1991;161:336. 13. Johnson MP, Bukowski TP, Kithier K, et al. Fetal urine albumin/globulin ratio in the in utero evaluation of obstructive uropathies [Abstract]. Am] Hum Genet 1992; 51:259.

Discussion K. SILVER, Evanston, Illinois. In contemplating the propostion of fetal therapy, at least four basic questions deserve consideration: (1) What to do (will what you do work being a subordinate issue)? (2) Who to do it to - all comers with a given diagnosis or a selected cohort meeting specific inclusion criteria? (3) What are the maternal and fetal hazards and are they quantifiable? (4) Will the perinatal outcome be improved relative to untreated cases? In the instance of isolated obstructive fetal uropathy we can dispense with the first question as having two possible answers, that is, percutaneous shunting and open fetal vesicostomy. Neither is known to be a superior approach, and the choice between alternatives is not the focus of this study. Although questions 3 and 4 were not stated objectives of thiS study, I would like to take advantage of this large clinical experience to consider both issues. The study'S methods have been considered by other investigators and rely on ultrasonographic imaging, rapid karyotyping, and fetal urinary index values obtained in a serial fa!lhion. By means of this database supplemented with patient counseling fetal renal prognosis, established in all cases, forms the basis from which a plan of management is constructed. One third of the reported pregnancies (n = 11) were electively terminated and 15 of the remaining 23 cases had shunting accomplished. The central question is whether the 100% success rate in predicting renal pathologic characteristics observed in this study can be duplicated prospectively? Before this is answered, some inconsistencies in the methods deserve comment. For example, a poor prognosis was anticipated in four continuing pregnancies, yet these fetuses underwent shunting anyway, a fact that is not consistent with the proposed algorithm. Did those patients insist on shunting in spite of the poor prognosis? Was gestational age or amniotic fluid volume a decision-making variable? Was "poor prognosis" classification less certain in prospect than after the fact? And why give equal weight to each urinary component (so that any single abnormality defined irreversible renal damage), when the individual components' ability to DR. RICHARD

June 1994 Am J Obstet Gynecol

predict renal dysplasia (expressed as negative predictive value in the study) ranged from 0.44 to 1.0? In interpreting the pathologic end point of renal dysplasia, it would also be helpful to know specific histologic information in each case, especially in the nonsurviving cohort, where poor renal function can only be inferred rather than measured. In how many of the poor-prognosis specimens was cortical tissue sparing seen? What was the degree of fibrosis in each case? Were the medullary pyramids always absent and was primary renal dysplasia or so-called nephronic dysgenesis ever suspected in any cases? This last concern recognizes the fact that renal dysplasia is a multifactorial entity with regard to cause and could, in selected cases, be independent of the obstructive uropathy. Such pathologic detail could also provide surrogate information about the potential for reversibility or preservation of impaired function. That seven of the 15 shunted fetuses did not experience long-term survival is an issue unrelated to the study'S objective, yet potentially of greater importance. Is this an improvement over what might have been expected in the preultrasonography era (question 4), and is this observation attributable in part to fetal hazard (question 3)? It is notable that fetal death, preterm delivery, and infection complicated a subset of these cases. Were more perinatal complications noted in cases with serial cystocenteses or in those requiring shunt replacement? Finally, on the basis of your experience, what percentage of fetuses would likely benefit from shunting (e.g., survive with normal renal function vs survive with renal insufficiency or die)? Could you estimate what study size would be required to explore such end points in a randomized fashion? DR. FEDERICO G. MAIuONA, Southfield, Michigan. Congenital anomalies represent one of the major causes of stillbirths and deaths in neonatal units. The development, evolution, and application of prenatal diagnosis have important implications for perinatal management, as this study implies. The evaluation and treatment protocol proposed in the paper has valuable points: (1) comprehensive fetal evaluation, including karyotype, and (2) serial fetal urinalysis, possible the mainstay for the use or nonuse of the cystoamniotic shunt. In addition, the inclusion of only low fetal urinary tract obstruction leading to bilateral, progressive deterioration in fetal renal function if left untreated must be emphasized. Earlier this year Lipitz et al.I reported their experience on the subject; they showed that 100% of the fetuses in whom they found abnormality elevated urinary levels of sodium and calcium showed histologic renal dysplasia on postmortem examination. Some of the vagaries of evaluating the fetal urine chemistry profile that we encounter in our own work in the early 1980s continues, as do the variations provided by fetal renal echogenicity on ultrasonographic examination. The authors emphasize their use of serial fetal urine

Johnson et al.

Volume 170, Number 6 Am J Obstet Gynecol

samples rather than a single tap as we did originally. The question then is how many amniofetocysto penetrations are needed to complete the "initial" evaluation before the true treatment starts? That count should include number and length of amnioinfusions, shunt insertions, and shunt replacements. Are parents made aware of this need and the possible use of antibiotics, immune globulin in cases of maternal Rh( -}, other possible isoimmunization dangers, prolonged bed rest, or even the use of tocolytics? The authors' ability to accurately diagnose fetal renal dysfunction (better than dysplasia, I believe) is coupled with their clear position "not to shunt" unless the comprehensive evaluation they propose rules out genetic or anatomic anomalies. Because they have evaluated fetuses as late as 24 weeks, can they give us an idea of how late have they initially shunted, because there are reasons to believe that the longer the total obstruction, the higher the system pressure and the worst the renal damage. 2 I would ask for a tighter definition of "early" total obstruction and how it relates to "short" and "moderate" duration. These aspects of the problem continue to present a clinical challenge to discuss with parents oligoanamnios created by prolonged premature rupture of membranes remote from term and pulmonary hypoplasia or limb deformation. The discussion on the pathophysiologic mechanisms of the renal damage is simple, short, and compatible with their philosophy for fetal cystoamniotic shunting, a term that I favor over "fetal surgery," which implies an extended, expanded access to fetal structures, rather than just the ultrasonic guidance of a trocar. Last, congratulations on what appears to be a 100% autopsy rate. The interventions and parental reactions in a setting of this type must be carefully evaluated and deserve a full report. Indeed, more questions come to mind to supplement the reasonable, straight answers provided in the paper. (I) Were the 11 women who selected pregnan0' termination fully "evaluated" before their decision to terminate the pregnancy? (2) Were all women fully "evaluated" before the counseling session to discuss follow-up therapy? (3) How was the nomogram for 132-microglobulinuria derived? What method was used for its determination? Did you perform serial examinations? How much does it cost? (4) How do you establish nonprocedure-related fetal loss? What percent loss did you quote parents? (5) Have you considered shunting with a "poor prognosis" evaluation at parental request? (6) Because patients were referred from institutions across the United States for this nontraditional therapy, can you give us an idea of the cost per case of a surviving neonate with full renal function? Please include any additional neonatal studies or procedures such as glomerular filtration rate, renal dialysis, surgeries, etc. (7) Because you have not included fetuses initially evaluated after 24 weeks, can you tell us the shunting versus gestational age at first insertion? Can you give us the

1777

shunting-to-delivery interval and delivery modality, so we have an overall outcome picture, especially for those women referred to their original physicians? REFERENCES 1. Lipitz S, Ryan G, Samuell C, et al. Fetal urine analysis for the assessment of renal function in obstructive uropathy. AM] OSSTET GYNECOL 1993;168:174-9. 2. Nicolini U, Tannirondom Y, Vaughn ], et al. Further predictors of renal dysplasia in fetal obstructive uropathy: bladder pressure and biochemistry of "fresh" urine. Prenat Diagn 1991;11:159-96.

DR. EMANUEL P. GAZIANO, Minneapolis, Minnesota. One of the frustrations in this area has been in terms of the catheter, not only its placement but its remaining functional; I want the authors' opinion of the Rodek catheter. Is it approved by the Food and Drug Administration? Is it experimental, and what kinds of reinsertion rates were necessary? Second, what is the earliest gestational age at which fetal urinary electrolytes are accurate and what is the earliest gestational age at which you've performed a catheter placement? DR. JOHNSON (Closing). Do we like the Rodeck shunt? Absolutely. It has taken 15 years to develop and has undergone a tremendous number of revisions; in my opinion it represents a technologic miracle as well as a mriosity . Is it approved by the Food and Drug Administration? No. This is experimental fetal therapy. The shunt is available now in the United States. Rocket of London, the manufacturer, now has an outlet on the East Coast; if you contact my office, I can provide you with that information. As to what gestational age the electrolytes are accurate, I believe that we are still collecting that data. It's very challenging to analyze underlying fetal renal function in the late first trimester and early second trimester; any of those of you who have been to the International Fetal Medicine and Surgery Society, of which I had the opportunity to chair the obstructive uropathy session last spring, know the discussion and arguments are tremendous and very entertaining. I believe that the data that we currently have show that we can evaluate underlying renal function in the fetus as early as 15 weeks' gestation. I will emphasize the point that was raised by Silver, that we do not give equal weight or value to each set of urinary electrolytes. We believe that the first vesicosentesis represents urine that has been somewhat stagnant and long-standing in the bladder. It does not truly reflect the underlying function or reserve of the fetal kidneys. By performing serial vesicocenteses we decompress and allow the drainage of the upper urinary tracts, therefore releasing the pressure intrarenally and allowing the fetal kidneys to tell us what they can do if the reflux pressure is taken off. So serial vesicocenteses are an integral part of our evaluation process. To answer Silver's other questions, there are a myriad

1778

Johnson et al.

of approaches to fetal obstructive uropathy that include vesicoamniotic shunting as we have described. There are Dr. Harrison's investigational protocols of taking the baby out of the uterus, performing a suprapubic vesicotomy, and replacing the fetus in utero. We are currently experimenting with other approaches using fine-needle embryo fetoscopy and looking in animal models and performing suprapubic vesicotomies in utero and resection of bladder neck obstructions from within the bladder using these fiberoptic systems. Why were patients with a poor prognosis shunted? We do not know all the answers at present. There is still some uncertainty, particularly when the urine values hover about the cutoff values; we truly believe in maintaining patient autonomy when we consider their requests to intervene when there is any doubt or uncertainty. As for the histologic characteristics in poor outcomes, we found significant underlying renal pathologic mechanisms, extensive fibrosis throughout the renal cortex and parenchyma. Some of these fetuses demonstrated early microcystic changes. Many of them showed well advanced microcystic and macrocystic dysplastic changes. Research that we are now performing with a fetal histopathologist shows that there's actually a good correlation between the degree of abnormalities in electrolytes and the extent and degree of damage that can then be seen histologically in the fetal kidneys. Were more perinatal complications seen in shunted fetuses? In the survivors that we have, who are now between 1.5 and 5.5 years old, there were prematurity issues in the immediate postnatal period as a result of the age at which they were delivered, which on average was about 35 weeks, with stabilization of these problems within the first 24 hours of life. The babies who had poor prognosis who were shunted and subsequently delivered had difficulties with renal function. Some of them had difficulties with pulmonary insufficiency secondary to the late gestational age at which the shunt was placed. How did we design a prospective study, and what is the size of the population? The international community, through the International Fetal Medicine Surgery Society, is a large collaborative of multinational groups of fetal researchers who are trying to put that data together. We collaborate closely with groups in Australia, New Zealand, England, Sweden, Italy, France, Germany; we are working together to collect these data so we can sort out this difficult problem. Were cases we have previously reported included in this cohort? Approximately 18 months ago we did present our initial experience in vesicoamniotic shunting. At that time we introduced the possibility or the concept of serial evaluations in examining these fetuses. This study follows up this concept by showing that serial vesicocentesis and analysis of renal function, if the pressure can be kept off the upper tracts, does, in fact, help in the analysis and evaluation of which fetuses will respond and benefit from vesicoamniotic shunting.

June 1994 Am J Obstet Gynecol

As for Dr. Mariona's extensive and long list of queries, "true treatment" we believe begins with the initial complete bladder drainage because it decompresses the upper renal tracts. It does reduce intrarenal pressures, which likely playa major role in inducing parenchymal damage. Our current recommendations are for a minimum of three urine samplings, which should establish a pattern of improving or worsening values and should provide the physician with some insight into the underlying renal function and perhaps renal reserve. A major focus of our program has been to completely involve the patients and their families in the evaluation process. We spend a tremendous amount of time and effort educating our patients and in continuous disclosure of our findings, interpretation of the data on the basis of our own and the combined world experience, repeated restatement of all available management options, and supporting these families in their decisionmaking processes, whatever they decide to do. No patient is ever coerced or directly counseled into our protocols, and participating patients may drop out of the evaluation sequence at any point without impact on the delivery of subsequent health care. A lengthy process of informed consent that includes the purpose, benefit, and limitations of this testing and a complete disclosure of potential risks for both the mother and the fetus obviously precedes all invasive procedures. Again, alternate forms of management are discussed repeatedly and are always available for the patient. Amnioinfusion is performed at the initial visit and isn't repeated unless necessary to facilitate the shunt placement. Also, in the past we had not routinely administered antibiotics at the time of shunt placement, although at present we have begun to do so on the basis of our recent study of amnioinfusion; RhoGAM (Rho[D] immune globulin) is routinely administered to all Rhnegative patients. The earliest shunt ever placed that we know of was at our facility at 14.5 weeks, and the latest shunt placement in this series was at 23 weeks. Our experience to date clearly shows that renal function in survivors who were shunted before 19 weeks' gestation was significantly better than those who were shunted after 19 weeks' gestation and further serves to emphasize the absolute need for early recognition, referral, and evaluation at centers performing this type of work and intervention, if appropriate, as early in the obstruction process as possible. We have observed three transient obstructions that are not included in these data. All of the cases in this study were complete outlet obstructions that presented with massive megacystis and upper system reflux and severe oligohydramnios, which to us indicates a longstanding process. These cases generally were diagnosed at 17, 18, or 19 weeks and immediately referred. Therefore we have no information as to when the obstruction process actually began, just evidence of a long-standing process.

Volume 170, Number 6 Am J Obstet Gynecol

As for fetal losses, we considered losses procedurerelated if they occurred within 1 week. One fetus was lost 10 days after shunt in a discordant twin pregnancy when premature rupture of membranes occurred in the opposite sac involving a completely normal fetus. All other losses occurred because of a variety of obstetrical problems at least 3 weeks after our surgical interventions. Although our experience is one of the largest in the world, we feel that no one really has sufficient data to quote specific loss rates. We therefore discuss our risks on the basis of our experience up front with the patient. Although over time we've become more confident in our diagnostic abilities, uncertainty still remains, particularly when urine values hover right around the cutoff values or just above, in what we consider the" gray zone." Because our predictive accuracy requires further experience, we do respect the patient's autonomy when we consider their requests for intervention. The average age of delivery was about 35 weeks, meaning that, because shunts were placed on average at 19 to 20 weeks, the pregnancies continued for an

Johnson at al.

1779

additional 15 to 16 weeks. Management of labor and route of delivery were not at all altered because the fetus had undergone this antenatal shunting. Neonatal evaluation was coordinated through our pediatric urology service and postnatally included determining the origin of obstruction, the degree of upper system reflux, and renal dynamics, including urine output. Neonatal interventions ranged from simple placement of a Foley catheter in posterior urethral valves to suprapubic vesicostomy. Because there are so few survivors and they're all still very young, I really cannot predict the cost per child of saving these infants from their otherwise dismal prognmis and inevitable fate. Last, 132-microglobulin values were determined by radioimmunoassays run in duplicate with triplex controls at a cost of about $8 to $9 per specimen. Cutoff values were determined on the basis of the reported experience of Dumez in Paris, with whom we collaborate through the International Fetal Medicine Surgery Society.