Accepte

Reprintnesi, GeAnkara,

0039-60

� 2013

http://d

A new and simple score for predictingcystobiliary fistula in patients withhepatic hydatid cystsBarıs Saylam, MD, Faruk Coskun, MD, Barıs Demiriz, MD, Veli Vural, MD, B€ulent Comcalı, MD, andMesut Tez, MD, Ankara, Turkey

Background. Hepatic hydatid cysts are common disorders in Turkey. Although most patients are treatedby percutaneous drainage, some cases require operative intervention. Biliary fistula is a majorcomplication of hydatid cyst operations. The purpose of this study is to identify preoperative predictors ofcystobiliary fistula (CBF) and to develop a scoring system for this disorder.Methods. Overall, 135 patients with hepatic hydatid cysts were included in this study. The followingvariables were analyzed as potential predictors of CBF: Age, gender, findings on physical examination,complete blood cell count, liver function tests, and ultrasonographic features of the cysts (type, diameter,number, and localization).Results. CBF was detected in 33 of 135 patients. Univariate analyses showed significant differences in cystdiameter, levels of alkaline phosphatase (ALP) and direct bilirubin, platelet count, and white blood cell(WBC) count between patients with and without CBF. On multivariate analyses, WBC count > 9,000/mm3 (odds ratio [OR], 4.5), direct bilirubin level> 0.7 mg/dL (OR, 2.76), cyst diameter> 8.2 cm (OR,5.48), and ALP level > 120 U/L (OR, 3.82) were significant and independent predictors of CBG. Onepoint was given for the presence of each of these factors to develop a new score. The resulting area under thereceiver operator characteristic curve was 0.803 (95% confidence interval, 0.726–0.866).Conclusion. Preoperative detection and management of CBF are important issues in the treatment ofhydatid cysts of the liver. Developing a scoring system based on routinely measured laboratory andradiologic factors will help the clinician to manage patients with hepatic hydatid cysts. External studiesare needed to validate this new scoring system in routine clinical practice. (Surgery 2013;153:699-704.)

From the Department of Surgery, Ankara Numune Training and Research Hospital, Ankara, Turkey

HEPATIC HYDATID CYSTS are believed to communicatewith the biliary system in #80% of patients.1 Earlydiagnosis and treatment are important for the suc-cessful management of cystobiliary fistulae (CBF).If untreated, CBF can cause a wide range of compli-cations, including biliary obstruction, cholangitis,and biliary cirrhosis.2 Communication betweenthe cyst and the biliary tree is classified as major orminor. Major (or frank) rupture is defined as cysticcontents draining directly into a major biliary ductcausing intermittent or complete obstruction ofthe duct. Major rupture occurs in 5–17% of patientsand is easily diagnosed preoperatively based on thepatient’s medical history and the results of physical

d for publication November 16, 2012.

requests: Faruk Coskun, MD, Ankara Numune Hasta-nel Cerrahi Klini�gi, Talatpasa Bulvarı 06100, Altında�g,Turkey. E-mail: farukcoskun@mynet.com.

60/$ - see front matter

Mosby, Inc. All rights reserved.

x.doi.org/10.1016/j.surg.2012.11.017

examinations and laboratory tests. Minor (occult orsimple) rupture is usually asymptomatic and can re-sult in postoperative biliocutaneous fistula or occultCBF in 13–37% of patients.3-5

The most common postoperative complicationis persistent external bile drainage, in which theresidual cyst communicates with the biliary tree.Although spontaneous closure may occur, persis-tent biliary fistula is not rare. The possibility ofidentifying patients at increased risk of CBF andwho are likely to develop late postoperative com-plications would allow us to initiate appropriatetreatments or interventions promptly that mightimprove the clinical outcomes.

Based on the results of the present and previousstudies, we propose a novel scoring system forhydatid cysts of the liver to provide surgeons with asimple, objective, and direct assessment of the riskof cystobiliary communication.

METHODS

We identified 152 patients with hepatic hydatiddisease who were admitted to the Surgical

SURGERY 699

SurgeryMay 2013

700 Saylam et al

Department at Ankara Numune Training and Re-search Hospital between 2000 and 2012. Data werecollected prospectively by dedicated audit officersfrom the hospital’s electronic and paper recordsand from the patients themselves. Thirteen patientswith increased preoperative levels of serum biliru-bin and liver function test were excluded from thestudy. These patients underwent preoperative en-doscopic retrograde cholangiography. One patientwith chronic renal failure, 2 with cardiac failure, and1 with severe chronic obstructive pulmonary diseasewere also excluded from the study. Most patientswere referred to our surgical clinic by the hospital’sinterventional radiology department after beingfound to be unsuitable for percutaneous aspiration;therefore, all patients were treated operatively.

A total of 135 patients, 45 men (33.3%) and 90women (66.7%), with amean age of 45 years (range,17–73) were included. All patients underwent con-ventional liver function tests (serum total and directbilirubin, aspartate aminotransferase, alanine ami-notransferase, g-glutamyl transpeptidase [GGT],and alkaline phosphatase [ALP]) and hematologicstudies. Plain radiography, abdominal ultrasonog-raphy, and, if necessary, computed tomographywere performed to evaluate the hydatid diseasestate. Cysts were classified radiologically as unilocu-lar (Gharbi types I and II), multilocular (Gharbitype III), or degenerate (Gharbi type IV). Thefollowing data were recorded for all 135 patients:Age, gender, laboratory data, cyst type, cyst size andlocation, presence of intraoperative CBF, type ofprocedure if CBF was present, type of surgery,presence of a postoperative biliary fistula, andtime to close the biliary fistula.

All operative procedures were conducted via anopen approach. The type of operative procedurewas determined by the individual surgeons. Allcysts were handled by unroofing of the cyst bypartial pericystectomy and cavity management.The type of cavity management, including externaldrainage, omentoplasty, or capitonnage, was at thesurgeon’s discretion. Hypertonic saline was used asthe scolicidal agent. At the start of the operation,all cysts were aspirated to shrink their volume anddecrease intracystic pressure. Hypertonic salinewas injected into the cyst cavity, and the cyst cavitywas opened 10 minutes later. After draining thecyst cavity, we placed a sponge into the cyst cavityand waited for a suitable time to detect bileleakage. If biliary leakage was detected, only pri-mary sutures or primary sutures with cholecystec-tomy and T-tube drainage were performed. Livertissue surrounding the periphery of the cyst wasexcised if deemed necessary. External drainage was

performed with a sump drain placed into the cystcavity and/or next to the cyst cavity. The drain wasremoved when the drainage stopped.

Independent variables. The following data werecollected and used in the analyses as independentvariables: Age, gender, results of liver function tests(aspartate aminotransferase, alanine aminotrans-ferase, ALP, GGT, and total and direct bilirubin),complete blood cell count, cyst location (right andleft or both), cyst diameter, cyst number, and cysttype (Gharbi type).

Dependent variable. The primary endpoint (de-pendent variable) was intraoperative or postoper-ative diagnosis of cystobiliary communication.

Statistical analysis. Continuous data are pre-sented as the mean values ± standard deviation.Differences in continuous variables were analyzedusing the Mann–Whitney U test. The Shapiro–Wilktest was used to assess normality. Categorical varia-bles were analyzed using Chi-square tests. Logisticregression was used to identify the factors associatedwith cystobiliary communication. Results of themultivariate analysis are shown as odds ratios (OR)with 95% confidence intervals (CI). Receiver opera-tor characteristic (ROC) curve analyses were used todetermine the optimal cutoff values for continuousvariables. A clinical score based on the final logisticregression model was constructed in which 1 pointwas assigned for the presence of each predictive fac-tor. Model discrimination was measured as the areaunder the ROC curve (AUC). The discrimination ofa prognosticmodel is considered perfect if AUC = 1,good if AUC is>0.8,moderate if AUC is 0.6–0.8, andpoor if AUC is <0.6.

RESULTS

A total of 192 cysts were treated in 135 patients.Most patients (n = 87; 64.4%) had 1 cyst; the remain-ing 48 patients (35.6%) hadmultiple cysts. The cystswere located in the right hepatic lobe in 83 patients(61.2%), the left lobe in 30 patients (22.4%), and inboth lobes in 22 patients (16.4%). The median cystdiameter was 7.5 ± 3.0 cm. Of the 135 patients, 90(67%) underwent partial cystectomy and omento-plasty, 35 (26%) underwent partial cystectomy andexternal drainage, and 10 (7%) underwent partialcystectomy. Bile leakage was detected in 33 patients(25%) intraoperatively or postoperatively. Thedemographic and clinical characteristics of the pa-tients with and without intraoperative/postopera-tive CBF are compared in Table I.

Univariate analyses. In univariate analyses, whiteblood cell (WBC) count, platelet count, direct bili-rubin level, cyst diameter, and ALP level were asso-ciated with a greater incidence of biliary leakage.

Table I. Demographic and clinical characteristics of patients with or without a cystobiliary fistula

VariableCystobiliary fistulapresent (n = 33)

Cystobiliary fistulaabsent (n = 102) P

Mean age (yrs) 41.4 ± 16.6 42 ± 15.1 NSGender, male/female 11/22 34/68 NSWhite blood cell count (per mm3) 10.3 ± 3.4 8.3 ± 3.2 .001Platelet count (per mm3) 364.150 ± 116.785 301.039 ± 88.587 .001Mean liver function test levels NS

Aspartate aminotransferase (U/L) 33.1 (10) 30.6 (27)Alanine aminotransferase (U/L) 33.8 (13) 34.9 (44)Alkaline phosphatase (U/L) 113.9 (31) 110.1 (32)g-Glutamyl transpeptidase (U/L) 93.3 ± 76.7 76.1 ± 55.5Direct bilirubin (mg/dL) 0.63 (0.36) 0.41 (0.28) .003

Cyst locationRight 19 64 NSLeft 6 24Right and left 8 14

Mean cyst diameter (cm) 9.8 ± 3.2 7.3 ± 2.7 .000Cyst type

Unilocular 15 43 NSMultilocular 11 37Degenerate 7 22

Surgery typePartial cystectomy and omentoplasty 20 70 NSPartial cystectomy and external drainage 24 11Partial cystectomy 2 8

Type of biliary interventionPrimary sutures 14 NSPrimary sutures with cholecystectomy 7T-tube drainage 12

Type of diseaseNew 30 81 NSRecurrent 3 21

NS, Not significant.

Table II. Multivariate logistic regression model forpredictors of the presence of a cystobiliary fistula

Regressioncoefficient

Oddsratio P

Scorepoints

Increased direct bilirubinlevel (>0.7 mg/dL)

1.01 2.76 .04 1

Increased white blood cellcount (>9,000/mm3)

1.27 4.5 .009 1

Cyst diameter (>8.2 cm) 1.7 5.48 .0006 1Increased alkalinephosphatase level(>120 U/L)

1.34 3.82 .01 1

SurgeryVolume 153, Number 5

Saylam et al 701

Multivariate risk prediction model and predic-tion score. All of the variables that could beassessed before operation were included in themultivariate model. Four variables were significantin this analysis: WBC count, >9,000/mm3 (OR,4.5); direct bilirubin level, >0.7 mg/dL (OR,2.76); cyst diameter, >8.2 cm (OR, 5.48); andALP level, >120 U/L (OR, 3.82; Table II). A prob-ability score was calculated by adding the numberof points assigned to each variable. Although theregression coefficients ranged from 1.01 to 1.7,for simplicity, 1 point was assigned to each of theserisk factors. The resulting NUMUNE (named afterour hospital) score (WBC count, cyst diameter, bil-irubin level, and ALP level) ranged from I to III.

Three groups of patients were defined based onthe NUMUNE score. The first group, with a scoreof I, comprised about 30% of the patients whoserisk of biliary leakage was <10%. The secondgroup included patients with a score of II, whohad a 10% risk of biliary leakage; this group

comprised of approximately 35% of the cohort.The third group, which comprised approximately35% of the patients, included those with a NU-MUNE score of III, whose risk of biliary leakagewas >50% (Table III).

The specificity, sensitivity, positive predictivevalue, negative predictive value, negative likelihood

score

0 20 40 60 80 1000

20

40

60

80

100

100-Specificity

ytivitisneS

Fig. ROC curve analysis of the NUMUNE score.

Table III. Risk of cystobiliary fistula according tothe NUMUNE score

NUMUNE score Number ofcystobiliary fistula

absent (%)

Number ofcystobiliary fistula

present (%)Score Points

I 0 38 (93) 3 (7)II 1 43 (90) 5 (10)III $2 21 (46) 25 (54)

SurgeryMay 2013

702 Saylam et al

ratio, and positive likelihood ratio for NUMUNEscores exceeding II were 79%, 76%, 54%, 91%, 0.31,and 3.68, respectively. The AUROC was 0.80 (95%CI, 0.73–0.87) for the NUMUNE score (Fig).

DISCUSSION

There is no consensus currently regarding theterminology to be used to describe hydatid cystswith CBF.6 We used the term CBF without consider-ing whether the patients had obstructive jaundice.Operative management of hepatic hydatid diseaseranges from radical procedures, such as hepatic re-section, to conservative procedures, such as cystevacuation. The aims of operative intervention inhydatid cyst disease are to inactivate the scolices,to prevent intraperitoneal spillage of cyst contents,to eliminate all viable elements of the cyst, and tomanage the residual cavity of the cyst, while mini-mizing complication.7

Preoperative detection and appropriate manage-ment of CBF are important factors in the treatment

of hepatic hydatid cysts, because such cysts areusually associated with increased risk of morbidityand mortality. In our study, identified the factorsassociated with the occurrence of CBF. Using theidentified factors, we then developed a new prog-nostic scoring system for CBF in patients withhepatic hydatid disease, which determine the NU-MUNE score.

We found no differences in age, gender, aspar-tate aminotransferase, alanine aminotransferase,GGT, and eosinophil count between patients withor without CBF. Notably, the type of cyst, primaryor recurrent status, the presence of single ormultiple cysts, and their location (right lobe, leftlobe, or both), did not affect the risk of CBF. Thesefindings are consistent with those of other studiesof CBF.1,2,8,11

Intracystic pressure increases with increasing cystdiameter.9 This hypothesis has been that, as the cystenlarges, it compresses the adjacent liver andstretches the bile channels in its immediate vicinity.Lateral openings develop in these over-stretchedducts, producing fistulae secondary to rupture ofthe hydatid.10 Demircan et al11 reported that acyst diameter of >8.5 cm was an independent pre-dictor of occult CBF. Atli et al8 found that cyst diam-eter of $14.5 cm was an independent predictor ofoccult intrabiliary rupture, although Kayaalp et al1

found that 65% of the cysts that caused biliary leak-age were <10 cm in diameter. Kilic et al7 reportedthat a cyst diameter of >7.5 cm was a risk factorfor intraoperative bile leakage and postoperativebiliary fistula. Unalp et al12 studied 183 cases withoccult and asymptomatic cysts and reported thatcysts of>10 cm in diameter predicted postoperativebiliary fistula. We also found that the diameter ofthe cyst was an independent predictor of CBF. Themean cyst diameter was 9.8 ± 3.2 cm in patientswith intrabiliary rupture versus 7.3 ± 2.7 cm in pa-tients without rupture (P < .001).

Increased intracystic pressure causing intermit-tent passage of cyst fluid, scolices, and minorfragments into the biliary system can cause in-creases in serum ALP, GGT, and bilirubin levels.8,11

Reabsorption of bile from the cyst cavity may alsobe responsible for the increased bilirubin andALP levels.8 Increased serum ALP and GGT levelswere reported to be markers for the communica-tion between hydatid cysts and bile ducts.1 Atliet al8 reported that increased serum ALP and di-rect bilirubin levels were risk factors for occultCBC. Atahan et al5 reported that GGT was greaterin patients with occult CBC and was more specificfor predicting occult CBF in patients with hepatichydatid disease compared with elevated ALP levels.

SurgeryVolume 153, Number 5

Saylam et al 703

Our findings support those of Atli et al, who re-ported increased risk of CBF in patients with in-creased serum ALP and direct bilirubin levels.

Platelets show cytotoxic activity in vitro andin vivo against extracellular parasites. This cyto-toxic process is mediated either by specific anti-bodies or directly through the parasite itself. Thesame particular metabolic pathways seem to beimplicated in both hemostasis and parasiticidalactivity.13 An earlier study evaluated platelet pa-rameters before and after the removal of hydatidcysts.14 Although the platelet count was not differ-ent between before and after removal, the meanplatelet volume and platelet mass were greater be-fore than after operation.14 Further studies areneeded to demonstrate the association betweenparasitic infestations and platelet parameters.

In their series of patients with occult cystobiliarycommunication, Unalp et al12 showed that serumALP level > 133 U/L, total bilirubin level > 1.2mg/dL, WBC count > 10,000/mm3, and cyst diam-eter > 10 cm were independent predictors of CBF.They also concluded that, if a combination ofthese 5 factors is present, the positive and negativepredictive values increased to 100% and 90%,respectively.12

The absence of a definition or a widely acceptedscoring system to predict CBF has hampered theability to interpret the outcomes. Accurate assess-ment and management of CBF are vital for appro-priate treatment and to conduct clinical trials.Preoperative recognition of patients at particularlyhigh risk for CBF may be useful for several reasons,including appropriate provision of informed con-sent, timing of operative intervention; access tohigher levels of care before and/or after opera-tion; and intra- and interdepartmental audit.

Using prospectively gathered data, we devel-oped and validated a novel scoring system topredict the risk of CBF in patients with hepatichydatid disease. Other researchers have identifiedseveral factors that are associated with an increasedrisk of CBF, but we believe we are the first to haveassessed these factors together. For a scoringsystem to be clinically useful, it must fulfill severalcriteria: It should use readily available and verifi-able clinical information; it should have beendeveloped and validated in the population inwhich it is to be used; and it should be free fromconfounding factors. The NUMUNE scoring sys-tem developed here uses data that are easilycollectable for any patient presenting with hepatichydatid disease. Serum ALP level > 120 U/L, WBCcount > 9,000/mm3, direct bilirubin level > 0.7mg/dL, and cyst diameter > 8.2 cm were

independent predictors for occult rupture in ourstudy. Using the results of multivariate logistic re-gression analysis, we developed the NUMUNEscoring system (score range, I–III) to predictCBF. A score of III had a sensitivity of 75%, a spec-ificity of 79%, and a positive predictive value of54% to predict CBF. ROC curves were plottedand the AUCs were calculated to compare the sen-sitivity and specificity for cutoff values. The result-ing AUC for the NUMUNE score was 0.803 (95%CI, 0.73–0.87).

The development of laparoscopic approaches totreat hepatic hydatid cysts has been introducedrecently in several centers. If bile leak is detectedwithin the cyst cavity during inspection, the areashould be drained and, if possible, secured withsutures to prevent further leakage. Conversion toan open operation may be necessary if laparo-scopic suturing is not possible.15 Bickel et al16

noted that bile peritonitis occurred in 18% of pa-tients where a postoperative drain was not usedand in no cases if a drain was used. We think thatour prediction score may prevent conversion toan open operation and bile peritonitis in laparo-scopic approaches.

In conclusion, we developed a simple scoringsystem (NUMUNE) using routinely collected radio-logic and laboratory parameters to predict the oc-currence of CBF. We suggest that preoperativeendoscopic retrograde cholangiography may beneeded to detect CBF in patients with a score of III.Validation studies of theNUMUNE score are neededto support its use in routine clinical practice.

REFERENCES

1. Kayaalp C, Bzeizi K, Demirbag AE, Akoglu M. Biliary com-plications after hydatid liver surgery: incidence and risk fac-tors. J Gastrointest Surg 2002;6:706-12.

2. Yıldırgan MI, Baso�glu M, Atamanalp SS, Aydinli B, Balik AA,Celebi F, et al. Intrabiliary rupture in liver hydatid cysts: re-sults of 20 years experience. Acta Chir Belg 2003;103:621-5.

3. Ozturk G, Yildirgan MI, Atamanalp SS, Basoglu M, AydinliB, Polat KY. An algorithm for the treatment of the biliarycomplications of hepatic hydatid disease. Turk J Med Sci2009;39:671-85.

4. Becker K, Frieling T, Saleh A, H€aussinger D. Resolution ofhydatid liver cyst by spontaneous rupture into the biliarytract. J Hepatol 1997;26:1408-12.

5. Atahan K, Kupeli H, Deniz M, G€ur S, C€okmez A, Tarcan E.Can occult cystobiliary fistulas in hepatic hydatid disease bepredicted before surgery? Int J Med Sci 2011;8:315-20.

6. Hamamci EO, Besim H, Sonisik M, Korkmaz A. Occult in-trabiliary rupture of hydatid cysts in the liver. World JSurg 2005;29:224-6.

7. Kilic M, Yoldas O, Koc M, Keskek M, Karakose N, Ertan T,et al. Can biliary-cyst communication be predicted beforesurgery for hepatic hydatid disease: does size matter? Am JSurg 2008;196:732-5.

SurgeryMay 2013

704 Saylam et al

8. Atli M, Kama NA, Yuksek YN, Doganay M, Gozalan U, Ko-looglu M, et al. Intrabiliary rupture of a hepatic hydatidcyst: associated clinical factors and proper management.Arch Surg 2001;136:1249-55.

9. Yalin R, Aktan AO, Yegen C, Dosluoglu HH. Significance ofintracystic pressure in abdominal hydatid disease. Br J Surg1992;79:1182-3.

10. Aarons BJ, Fitzpatrick SC. Hepato-biliary hydatid disease.In: Kune GA, Sali A, editors. The practice of biliary surgery.Oxford: Blackwell Scientific Publication; 1980. p. 399-416.

11. Demircan O, Baymus M, Seydaoglu G, Akinoglu A, SakmanG. Occult cystobiliary communication presenting as postop-erative biliary leakage after hydatid liver surgery: are theresignificant preoperative clinical predictors? Can J Surg2006;49:177-84.

“Well, here’s the plan. Fipractices, replace the docand then move to patientscomputers. Think of the p

12. Unalp HR, Baydar B, Kamer E, Yilmaz Y, Issever H, TarcanE. Asymptomatic occult cysto-biliary communication with-out bile into cavity of the liver hydatid cyst: a pitfall in con-servative surgery. Int J Surg 2009;7:387-91.

13. Polack B, Peyron F, Auriault C. Platelet cytotoxicity againstparasites. Nouv Rev Fr Hematol 1991;33:317-22.

14. Kucukbayrak A, Oz G, Fındık G, Karaoglanoglu N, Kaya S,et al. Evaluation of platelet parameters in patients with pul-monary hydatid cyst. Mediterr J Hematol Infect Dis 2010;2:2035-6.

15. Koea JB. Laparoscopic treatment of hepatic hydatid disease.ANZ J Surg 2012;82:499-504.

16. Bickel A, Loberant N, Singer-Jordan J, Goldfeld M, Daud G,Eitan A. The laparoscopic approach to abdominal hydatidcysts. Arch Surg 2001;136:789-95.

rst we buy the primary care s with nurse practitioners, just interacting with rofits!!”