Splenopancreatic DisconnectionImproved Selectivity ofDistal Splenorenal Shunt

W. DEAN WARREN, M.D. KAREEM M. ABU-ELMAGD, M.B., B.CH. WILLIAM 0. RICHARDS, M.D.WILLIAM J. MILLIKAN, JR., M.D. JOHN R. GALLOWAY, M.D. ATEF A. SALAM, M.D.J. MICHAEL HENDERSON, F.R.C.S. G. TOM SHIRES III, M.D. MICHAEL H. KUTNER, PH.D.

Distal splenorenal shunt (DSRS) improves survival from varicealbleeding in nonalcoholic cirrhotics but not in alcoholic subjects.The metabolic response after DSRS is also different in alcoholicand nonalcoholic cirrhotics. Portal perfusion, quality of bloodperfusing the liver, cardiac output, and liver blood flow do notchange in nonalcoholics. In alcoholics, portal perfusion is fre-quently lost (60%), quality of blood perfusing the liver decreases,and cardiac output and liver blood flow increase. It is proposedthat portal flow is lost in alcoholics via pancreatic and coloniccollaterals after surgery. Elimination of this sump by addingcomplete dissection of the splenic vein and division of the splen-ocolic ligament to DSRS (splenopancreatic disconnection, SPD)could preserve portal perfusion, decrease shunt loss of hepato-trophic factor, and improve survival in alcoholic cirrhotics. Thisreport compares data 1 year after surgery in two groups of cir-rhotics: group 1 (8 nonalcoholic; 16 alcoholic) had DSRS withoutSPD; group II (17 nonalcoholic; 11 alcoholic) received DSRS +SPD. Methods: Portal perfusion grade, cardiac output (CO),liver blood flow (f), hepatic function (GEC), and hepatic volume(vol) were measured before and 1 year after surgery. Shunt lossof hepatotrophic factor was estimated by insulin response (changein plasma concentration over 10 minutes: AUC) after argininestimulation. Results: Groups I and II were similar before surgery.Metabolically, nonalcoholics remained stable after both DSRSand DSRS + SPD. After standard DSRS, alcoholics lost portalperfusion (75%, p < 0.05), CO, and f increased (p < 0.05), andquality of blood perfusing the liver was decreased (GEC/f: p<0.05). DSRS + SPD preserved portal perfusion better (p <0.05) in alcoholic cirrhotics than did DSRS alone. After DSRS+ SPD, the metabolic response in alcoholics resembled that ofnonalcoholics. CO, f, and GEC/f remained stable. These datashow: (1) DSRS + SPD preserves postoperative portal perfusionin alcoholic cirrhotics better than DSRS alone. (2) Metabolicresponse to DSRS + SPD is similar in alcoholic and nonalcoholiccirrhotics. (3) Because portal perfusion and metabolic integrity

Presented at the 106th Annual Meeting of the American Surgical As-sociation, Hot Springs, Virginia, April 24-26, 1986.

Supported in part by United States Public Service Grant AM 15736.Reprint requests: W. Dean Warren, M.D., Joseph B. Whitehead Pro-

fessor and Chairman, Department of Surgery, 1364 Clifton Road, N.E.,Atlanta, GA 30322.

Submitted for publication: April 28, 1986.

From the Departments of Surgery and Biometry, EmoryUniversity School of Medicine, Emory University Hospital,

Atlanta, Georgia

are preserved after DSRS + SPD, its use in alcoholic cirrhoticsshould improve survival.

D-v URING THE LAST 5 YEARS, endoscopic sclero-

therapy has re-emerged as the safest method totreat acute bleeding from gastroesophageal var-

ices."' However, there is a significant failure rate aftersclerotherapy,"5"11 and surgery remains the only definitiveway to prevent rebleeding. In nonalcoholic cirrhotics andpatients with extrahepatic portal vein thrombosis or pri-mary hepatic fibrosis, selective variceal decompression bydistal splenorenal shunt (DSRS) represents optimal sur-gical therapy because postoperative portal perfusion ismaintained for years and survival is improved.'2-'6 In al-coholic cirrhotics, however, portal perfusion is lost in 60%of patients 1 year after DSRS, and survival is no betterthan that achieved by total portasystemic shunt.'5"16

In 1984, Warren proposed that loss ofportal perfusionafter selective shunt occurred via the system of transpan-creatic and colonic collaterals that developed after selectiveshunt between the high pressure portal circulation andthe low pressure splenorenal anastomosis.'4"7 Loss ofhepatotrophic factors (insulin) from the liver through this"pancreatic siphon" (Fig. 1)18 was offered as one reasonwhy selective shunt did not prolong survival in alcoholicswho lost portal perfusion. Warren projected that, if thesplenic vein were completely separated from the pancreasduring selective shunt (splenopancreatic disconnection,SPD) (Fig. 2), the pancreatic siphon would not developand postoperative portal perfusion and the selectivity ofthe operation could be preserved. Shortly after Warren'sobservations were published, Inokuchi described similar

346

SPLENOPANCREATIC DISCONNECTION

collateralization after selective shunt ("portal malcircu-lation") and corroborated its physiologic significance tothe loss of portal perfusion.'9

This report presents metabolic and hemodynamic data1 year after surgery in a group of cirrhotics operated withDSRS plus SPD. For purposes of comparison, data froman unmatched but similar group ofpatients who receivedselective shunt without splenopancreatic disconnectionare also presented.'7

Material and Methods

Technical Aspects ofDSRS + SPD

The patient is positioned with the left side elevated 20degrees and the left arm adducted across the chest. A leftsubcostal incision is extended across the right rectus mus-cle. After dividing the gastrocolic ligament from the py-lorus to the short gastric vessels, the splenocolic ligamentis taken down, which exposes the hilum of the spleen,provides access to the lesser sac, splenic artery, and tailof the pancreas, and interrupts the major collateral path-way between the splenic vein and mesentery ofthe splenicflexure.SPD is complete dissection ofthe splenic vein from the

pancreas to its bifurcation at the splenic hilum. Dissectionof the splenic vein should proceed on the vein with theposterior, inferior surface isolated before anterior and su-perior surfaces are approached.20 The junction of thesplenic and superior mesenteric vein should be controlledearly in the dissection. Small splenopancreatic perforatingveins are ligated and divided as they are encountered.Division of the splenic vein at its junction with the su-perior mesenteric vein will facilitate dissection of thesplenic vein from the pancreas, as does intermittentclamping of the splenic artery to decrease blood loss. Su-periorly rotating the pancreas with the surgeon's left handbehind the gland aids exposure as the splenic vein passesthrough the upper border of the pancreas (the pancreaticgroove). In most cases, the splenic vein courses above thepancreas for several centimeters before entering the hilumof the spleen. Controlling the vein above the pancreasrequires special care, since injury to the vein at the hilummay cause irreversible harm. If dissection of the splenicvein from the pancreas is difficult, complete mobilizationof the spleen from its bed with rotation of both spleenand pancreas medially will allow dissection ofthe splenicvein from a posterior approach (Inokuchi maneuver).'9The splenorenal anastomosis is performed in the usual

manner, as is interruption of the left and right gastricveins.20

Patient Populations

Selective decompression was performed in 111 patientsbetween August 1983 and January 1986. Sixty-five pa-

Inf. mesentericv. ligated

FIG. 1. Schematic of the "pancreatic siphon" that develops after DSRS,resulting in the loss of hepatotrophic factors from the liver. Arrows denoteroute of blood flow from the high pressure portal circulation to the lowpressure splenorenal anastomosis.

tients had DSRS + SPD, and the remaining 46 patientsreceived standard DSRS. The latter was usually performedin the emergent or urgent setting. Twenty-eight of theDSRS + SPD patients (17 nonalcoholic; 11 alcoholic)have been followed at least 1 year and define the studypopulation.An additional 10 patients with DSRS + SPD and 13

others who had standard DSRS had stimulated levels ofinsulin measured in their shunts at a median follow-upof 5 months (range: 3-15 months; X = 6.7 months) and31 months (range: 6-60 months; X = 33.8 months), re-spectively.

Tall of pancreas

Coronary v6ln /s

Portal vein

stump

Supefior ~~ ~ ~ ~ ~ ~ ~~~Slenc veinmesenteric attached tovein L renal vein

IVC! \ ~Gonadal vein

InPerior mesentericvein lited

FIG. 2. DSRS + SPD. Note the complete separation of the splenic veinfrom the pancreas.

347Vod. 204 * No. 4

Ann. Surg. * October 1986

A previously published group'7 of 24 cirrhotics (16 al-coholic; 8 nonalcoholic) are included to compare the he-modynamic and metabolic differences between DSRS andDSRS + SPD.The resurgence of endoscopic sclerosis has changed our

surgical population. Most (80%) patients now operatedon are sclerotherapy failures and are brought to surgeryunder less than ideal conditions. These patients are sicker,as indicated by the larger population ofClass C subjects"and the altered pattern of their quantitative tests beforeoperation (see Tables 2 and 3).

Evaluation Methods (Table 1)

Patients were evaluated before and after surgery withthe clinical, biochemical, and hematologic studies thatcomprise the Emory Hepatic Database.2' Liver biopsyconfirmed the etiology of the cirrhosis.

Hepatic function was quantitated by galactose elimi-nation capacity (GEC).22 Liver blood flow was estimatedby low dose galactose clearance (flow).23 Cardiac outputwas measured by echocardiography or first pass nuclearcardiography (CO).25 Portal perfusion was graded on thevenous phase of the superior mesenteric angiogram: 1(normal) -- 4 (absent or reversal of flow).26 Liver volumewas computed from abdominal computed tomographicscan (vol).24

Quality of blood perfusing the liver was indexed byliver function divided by liver blood flow (GEC/flow).'7

Insulin Studies

Loss of pancreatic hepatotrophic factor (insulin)through the shunt was estimated by measuring plasmainsulin concentrations after intravenous arginine stimu-lation (50 ml of 10% solution/90 sec). Blood samples weredrawn through a catheter placed via the femoral and leftrenal veins 2 cm into the splenorenal shunt during an-giography. Simultaneous peripheral venous samples weredrawn from the antecubital vein opposite the site of theinfusion. Samples were collected before and at 2, 4, 6,and 10 minutes after arginine bolus. Insulin was measuredby radioimmunoassay27 and response defined as the areaunder the plasma concentration versus time curve (0-10minutes).

Statistical Methods

Pre- to 1 year postoperative changes were comparedusing paired t-test.28

Results

Clinical

Splenopancreatic disconnection was performed in morethan 90% ofpatients in whom it was attempted. The extra

dissection extended surgery 30-60 minutes and increasedtransfusion requirements in some patients. Separating theentire length of the splenic vein from the pancreas wasmore difficult in alcoholic patients and subjects wherechronic sclerotherapy had produced perivenular thick-ening and peripancreatic inflammation.

Operative mortality (in hospital or 30 days) after DSRS+ SPD was 8% (5 of 65 patients). Four patients died ofhepatic failure, which was linked to sepsis in three. Thefifth patient died of complications related to intra-ab-dominal bleeding.Two patients (3%) bled from recurrent varices in the

early postoperative period and required portacaval shunt.Two others (3%) had intra-abdominal bleeding which re-quired re-exploration. This incidence of rebleeding is nothigher than after standard DSRS.' 5"6 It had been antici-pated that the extra dissection and subjection ofthe pan-creas to undecompressed portal hypertensive pressureswould increase postoperative bleeding.

Quantitative

The pre- to 1 year postoperative data for the 28 patientsoperated with DSRS + SPD is presented in Table 1.Comparison of data from the DSRS + SPD group withpatients receiving DSRS without SPD is listed in Tables2 and 3 and portrayed in Figures 3-8.

Portal perfusion. Portal perfusion was preserved innonalcoholics and lost in alcoholics 1 year after DSRSwithout SPD.'7 Portal perfusion was, preserved in bothalcoholics and nonalcoholics 1 year after DSRS + SPD(Table 1). DSRS + SPD preserved portal perfusion better(p < 0.05) than DSRS without SPD in the alcoholic subset(Fig. 3).

Cardiac output. Cardiac output increased (p < 0.05) 1year after DSRS without SPD when portal perfusion waslost in the alcoholic subset (Fig. 4). Cardiac output didnot increase in nonalcoholics after standard DSRS whenportal perfusion was preserved (Table 3). Cardiac outputdid not increase in alcoholics or nonalcoholics operatedon with DSRS + SPD. These changes in cardiac outputwere significantly different (p < 0.05) in alcoholics op-erated on with DSRS + SPD compared to alcoholics re-ceiving DSRS without SPD (Fig. 4).

Liver bloodflow. Liver blood flow increased (p < 0.05)in alcoholic cirrhotics after standard DSRS 1 year aftersurgery (Fig. 5, Table 3). 7 Liver blood flow decreased (p< 0.05) in alcoholic cirrhotics 1 year after DSRS + SPD.Liver blood flow was higher (p < 0.05) before surgery inalcoholics who received DSRS + SPD. The difference be-tween the change in liver blood flow (LBF, DSRS + SPD:postoperative-preoperative vs. LBF, DSRS without SPD:postoperative-preoperative) was also significant (p< 0.05) (Fig. 5). Liver blood flow did not increase after

348 WARREN AND OTHERS

SPLENOPANCREATIC DISCONNECTION

N_U _- -0 o00- n t-N

C 0% Cl 00 %0 %000 %n

0% C 0N o oo I-*N _

-0%~I000000r -, r

Cl --C -I' c7 tn IC I'* o ot

0% N °s - t 0%\0000-e

0000 'IOC~ 0nin

" en "o m m

0- 0%4 O.- 0-- >0 0Cl Ol' 4I-4

- 0-- l - C - -

-etrm Clr ~ Co o

-N 'oNoo^ 0fi

- -- - NN- -- -

0,-, n VI) N- all N ) 0%, 00cl Cl cl 0)NO - 0 Cl4'

Rt O l- - l-0% ON0O

- Cl 0 -C 0-0- Clo

^ tt ^ ntnd-w

'Im Cl m - m 0% 0t00O . N A . 0 0 C

1% tn 'I lqo o1v0 " m

ci oo r- o " -t oo t oo Ci Ci en _- rIZ r; r- wo r oo r- ,6 oo -ti r .6 c cy r

IC 0O

N\ en vO~ oOCh1'I Cl

ol~r-

O- 00 - -t " - - a,Nd-N

o 0Cl ".Q eM e - M .0 1.O 'in Nl 000C-1-0 t N0% 0 0 NIC 0% - C -

N 0% 'I

~. If~ s~ 0000Cl 000- ~ rcN-Cl.rn-o£N0%C^ ^00% Cl xor 0 N o0

-CONl---r- a.

-

000e"M -.0 nf 0n0-0 -n

C t m oC r0000- C

IM0. 0% O0N10nI Cl 00 %. 0 00,00-lr - -lC -~ -~- C--- -lo00 0o 0o= IC '0-

-0 l oO oo Nr.'o 00-t.t. -o 0 w 'Iv

Clro Cl CletO ClN Clm Cl -

.}xooo- t.n.oOt

\0 NoO ^£ e o o tO>

O-Cl0.-i Cl -000-0 --Cl

elO-OOO-OOO0C--^0

Cl 0% NN Cl C0C 0- 0% I00 Cl N

cl rCl4 .rlelicl el; r-i el el; c.0 fli 'li C0 el;

.~~~~~~ .- .- ..- .

. . . . . g. . . [. . . . .

Noooo~q- ON -o- (ON W - r-

IR o oo4 n1. R

C.)

C.) 0._

'a u

o-ClCt.r)0N000%0- =- Rt Coo=ON-.) - -0

6 z

Vol. 204 * No. 4

Cu

Cu

349

00

00

0 +1 g

C.)

_ 0

0) _

+I )

0

ene

.0

C._0)

C.)

C.)

0D

.

U

U)

*U)

Cu

0.

C.)

CuC.)*a0

C.)

UD

00

0-o

w

Cu

0)C-Cd

WARREN AND OTHERS Ann. Sur& * October 1986

TABLE 2. Biochemical Changes before and at 1 Year after DSRS and DSRS+SPD

Total Alcoholic Nonalcoholic

Preop I Year Preop I Year Preop I Year

Bilirubin (0.1- 1.1 mg/dl)DSRS+SPD 1.4±0.9 2.3 ± 3.2 1.5 ±0.8 1.4±0.8 1.4± 1.0 2.7 ±4.0DSRS 1.1 ±0.5 2.0±0.9 1.0±0.4 2.1±1.0 1.1 ±0.5 1.8±0.7

Albumin (3.2-5.6 mg/dl)DSRS+SPD 3.4 ± 0.5 3.5 ± 0.7 3.4 ± 0.5 3.8 ± 0.5 3.4 ± 0.6 3.3 ± 0.8DSRS 3.8 ± 0.5 3.6 ± 0.4 3.6 ± 0.5 3.6 ± 0.4 3.9 ± 0.4 3.5 ± 0.5

Prothrombin time (sec. prolonged)DSRS+SPD 1.2 ±0.9 1.0±0.8 1.4±0.9 1.1 ±0.8 1.1± 0.8 0.9 ±0.8DSRS 2.2 ± 1.0 1.6 ± 1.0 2.6 ± 0.9 1.7 ±0.9 1.8 ± 1.0 1.5 ± 1.3

either operation in patients with nonalcoholic cirrhosis(Table 3).

Quantitative liverfunction (Tables I and 3). Galactoseelimination capacity did not decrease significantly inany of the patient populations after DSRS with or with-out SPD.

TABLE 3. Function and Hemodynamic ChangesPreoperative and at I yr

DSRS DSRS + SPD(Mean ± S.D.) (Mean ± S.D.)

Preop 1 Year Preop I Year

GEC (500 ± 50mg/min)

Total 350±99 315±81 323±66 316±66Alcoholics 337 ± 99 305 ± 69 313 ± 56 305 ±65Nonalcoholics 326 ± 98 324 ± 93 327 ± 72 320 ± 73

Flow (ml/min)Total 1089 ± 267 1152 ± 288 1399 ± 483 1207 ± 347Alcoholics 1133 ± 265 1339 ± 406 1717 ± 545 1297 ± 346Nonalcoholics 1045 ± 269 964 ±169 1203 ±300 1102 ± 321

GEC/fow X 100Total 32.1 ±7.9 27.8 ± 8.1 25.8 ±9.0 28.9 ± 1.0Alcoholics 30.0 ± 7.3 24.7 ± 9.2 20.0 ± 9.0 23.4 ± 6.4Nonalcoholics 35.2 ± 7.8 33.5 ± 6.6 30.1 ± 7.6 32.3 ± 10.4

Volume1(1493 ± 2309cm3)

Total 1801 ±517 1574±562 1518±523 1403±52Alcoholics 2113±600 1836±637 1827±520 1677±602Nonalcoholics 1489 ±433 1311± 487 1306 ±431 1216 ±464

GEC/volumeX 100

Total 21.0 ± 6.0 21.2 ± 5.5 24.6 ± 9.9 25.1 ± 8.8Alcoholics 17.3 ± 7.0 18.5 ± 6.6 19.5 ± 9.4 20.9 ± 8.3Nonalcoholics 24.6 ± 4.1 25.6 ± 4.9 27.2 ± 9.4 28.3 ± 8.3

Cardiac output(3.5-5.5L/min)

Total 6.9 ± 1.5 9.3 ± 2.9 8.3 ± 2.8 7.9 ± 1.9Alcoholics 6.8 ± 2.0 10.7 ± 3.0 8.3 ± 3.2 8.3 ± 2.6Nonalcoholics 6.9 ± 1.0 7.9 ± 2.9 8.3 ± 2.5 7.2 ± 1.1

Quantitative liverfunction/liver bloodflow. GEC/unitflow indexes hepatic function for a given volume of flowand is a measure of the quality of blood perfusing theliver (Fig. 6, Tables 1 and 3).17 GEC/unit flow decreased(p < 0.05) in alcoholics who lost portal perfusion afterDSRS without SPD. This resulted from a decrease in GECand increase in liver blood flow (p < 0.05). When portalperfusion was preserved in alcoholics after DSRS + SPD,GEC decrease was less and liver blood flow also decreased.The difference in change in GEC/flow (GEC/f: DSRS+ SPD, postoperative-preoperative vs. DSRS withoutSPD, postoperative-preoperative) was significantly dif-ferent (p < 0.05) (Fig. 6, Table 3). GEC/f did not changein nonalcoholic cirrhotics after either operation whenportal perfusion was preserved.

Liver volume. Liver volume decreased (p < 0.05) in allpatients after both operations (Fig. 7, Tables 1 and 3). 17

Similar changes in liver volume have been observed inunoperated cirrhotics treated with chronic sclerother-apy.'"3 Liver function/volume (GEC/vol X( 100) was

lower (p < 0.05) before surgery in alcoholic cirrhotics,

.O 20

aC 3

-a

0 41

DSRS

1

2-

3-

4-

Pre-Op One Year

DSRS + SPD

Pre-Op One Year

FIG. 3. Portal perfusion before and after operation in alcoholic patients.One represents normal perfusion; 2 and 3 are intermediate grades ofperfusion; 4 is absent perfusion. Perfusion is better maintained at 1 yearin the DSRS + SPD group.

350

SPLENOPANCREATIC DISCONNECTION

DSRS

p<0.05

14'

12'

10'

8-

6'

4'

2'

Pre-Op One Year

DSRS + SPD50"

40"00

30"

E 20"

E10

Pre-Op One Year

DSRS 50'

40'

30'

20

. p<0.05 10

Pre-Op One Year

DSRS + SPD

Pre-Op One Year

FIG. 4. Pre- and postoperative cardiac outputs in alcoholic patients (X± S.E.). Cardiac output is not increased at I year following DSRS + SPD.

patients with alcoholic cirrhosis having selective shunthad larger livers. This index (GEC/vol X 100) did notchange significantly in alcoholics or nonalcoholics aftereither operation.

Insulin studies (Fig. 8). Peripheral and shunt insulinlevels rose (p < 0.05) after arginine stimulation. Shuntlevels of insulin were greater (p < 0.05) in patients witha pancreatic siphon after DSRS without SPD at 2 and 4minutes after arginine stimulation than in patients withouta pancreatic siphon after DSRS + SPD. Total loss of in-sulin over 10 minutes was also greater (p < 0.05) (Fig. 8)in patients who had a pancreatic siphon after DSRSwithout SPD compared to patients without a pancreaticsiphon (AUC).

Discussion

SPD was conceived and implemented as a technicaladdition to DSRS to preserve portal perfusion and main-tain the selectivity of the operation. Rikkers was first todocument that nonalcoholics maintained portal perfusionbetter than alcoholics after selective shunt. 4 Zeppa madethe crucial observation that survival was better in non-alcoholics than alcoholic cirrhotics after selective shunt.31Henderson showed that when portal perfusion was lost

FIG. 6. Liver function per unit flow (GEC/flow x 100) before and afteroperation in alcoholic patients (X ± S.E.). Lower values indicate thatan increasedblood flow is needed to maintain a specific level ofhepatocytefunction. The quality of blood perfusing the liver decreased followingDSRS while remaining unchanged at 1 year following DSRS + SPD.

after selective shunt in alcoholics, the hyperdynamic statecharacterized by an increase in cardiac output and liverblood flow also developed. Henderson postulated that theloss of portal perfusion and the hyperdynamic state werelinked to decreased survival after selective shunt in thealcoholic.'7 Other data reinforced this concept. Althoughquantitative liver function was maintained 1 year afterselective shunt when portal perfusion was lost,'7 longerfollow-up showed that liver function decreased and theincidence ofhepatic encephalopathy increased in patientswithout prograde portal flow.'4 The 10-year follow-up ofEmory's controlled trial reinforced the importance ofmaintaining portal perfusion over time.'5 The only pa-tients with patent shunts who maintained hepatic functionand encephalopathy-free existence were those in whomportal perfusion was preserved.'5

These observations led to the concept ofthe pancreaticsiphon (Fig. 1) and the addition of the SPD to selectiveshunt. DSRS + SPD eliminates the pancreatic siphon andpreserves postoperative portal perfusion in the alcoholiccirrhotic better (p < 0.05) than DSRS without SPD. These

DSRS

l p40.05

2000-

1800-

1600-

1400'

1200'

1000-

onn.-

Pre-Op One Year

DSRS + SPD

* p<0.05

Pre-Op One Year

FIG. 5. Liver blood flow before and after operation in alcoholic patients(X ± S.E.). Liver blood flow significantly increased at 1 year after DSRSwhile falling after DSRS + SPD.

2800"

2400"

2000"

'El 1600"

12001

DSRS2800"

2400"

2000"1600"

. p<0.05 1200"

Pre-Op One Year

DSRS + SPD

.

p40.05

Pre-Op One Year

FIG. 7. Liver volume before and after operation in alcoholic patients (X± S.E.). Both DSRS and DSRS + SPD show a decrease in liver volumeat I year, and there is no difference in the rate of decrease between thetwo operations.

14

12'

10.C.EE 8'I.co0 6'

4-

2'nt

351

2000'

1800

1600'.E,E 1400-

1200.

1000'

Vol. 204 * No. 4

352400360320'

- 280

v. 240:

c200'

1' 160cn

'- 1208040n

WARREN AND OTHERS

DSRS 969±22/ DSRS + SPD 360±12

J... n

Arginine Infusion

0 2 4 6Time (minutes)

3?7

T,-To

304±62119+24

p<0.05

8 10

FIG. 8. Shunt insulin levels following stimulation of the pancreas byarginine infusion (X ± S.E.). Loss of insulin through the splenorenalshunt is greater after DSRS, as denoted by the large area under the insulinconcentration curve (AUC).

data are supported by clinical experiments published byInokuchi.'9 In his group of nonalcoholic subjects, Inok-uchi emphasized the importance of minimizing the lengthof splenic vein left in the pancreas in communication withthe portal vein. Even a short segment of splenic vein pro-vides the origin for the development of an extensive net-work of collaterals that can-over time-enlarge and si-phon flow from the high pressure portal circulation to thelow pressure shunt. We concur with Inokuchi's conceptand feel that the most efficacious approach is to harvestthe entire splenic vein from its junction with the superiormesenteric vein to the splenic hilum. The Italian group32has also presented experience with a modified SPD thatreinforces data presented in this report.The data in this study also show that concurrent with

preservation of portal perfusion after DSRS + SPD, thehyperdynamic state does not evolve in the alcoholic cir-rhotic. The absence of the pancreatic pathway appears toprotect against the development of a chronic hyperdyn-amic state. All patients after both DSRS and DSRS + SPDshow a mild hyperdynamic response early in the post-operative period, which disappears after several monthsin patients who maintain portal flow (unpublished data).The early development ofa low resistance collateral path-way (through the pancreas) during this phase may be crit-ical to both loss of portal flow and the persistence of thehyperdynamic state. These possibilities remain conjec-tural. The data from this study show good maintenanceof portal flow, no hyperdynamic response, and a favorablefunction/unit blood flow ratio in the alcoholic population.The hemodynamic and metabolic response to DSRS+ SPD in alcoholics is similar to that seen after DSRS innonalcoholics and unoperated patients treated withchronic sclerotherapy. Further work is required to definethe mechanisms by which this has been achieved.

These data also show that when portal perfusion is pre-served after DSRS + SPD, less insulin is lost through the

Ann. Surg. * October 1986

shunt. This study does not show that more insulin wasdelivered to the liver. Further studies must be done toshow if this occurs and to determine ifhepatocyte integrityis enhanced.33 This will require direct sampling of portalblood.

This report has focused on the alcoholic subset becausethis group loses portal perfusion after standard selectiveshunt. Should selective shunt plus disconnection be per-formed in nonalcoholics whose portal perfusion and liverfunction is already preserved long-term by standard shunt?We believe so because the pancreatic siphon also developsin nonalcoholics. Indeed, it was the loss of liver volumein the nonalcoholic (whose change in liver size is not sub-ject to bouts ofalcoholic hepatitis) that directed attentionto the pancreatic siphon. We now know that loss of livervolume occurs in unoperated patients"' 30 and probablyrepresents the natural history of disease. The very im-portant work of Professor Inokuchi proves the value ofseparating splenic vein collaterals from the portal circu-lation in the nonalcoholic cirrhotic.'9The selective shunt plus splenopancreatic disconnection

controls rebleeding as well as standard distal splenorenalshunt. The results ofthis initial experience show that onlytwo patients with patent shunts rebled from varices (3%).One additional patient-one ofthe hospital mortalities-also bled from varices after disconnection, but bleedingwas associated with terminal liver failure. We believe,however, that variceal bleeding after disconnection is he-modynamically different from variceal bleeding afterstandard selective shunt. After standard selective shunt,renal vein hypertension is common (10%) and constitutesthe major cause of early rebleeding.'6 Complete dissectionof the splenic vein from the pancreas eliminates thesplenopancreatic tributaries and collaterals between theretroperitoneum and splenic vein that may aid varicealdecompression. Hence the full dissection, coupled withgastric devascularization and coronary azygous discon-nection, leaves the short gastric veins as the only pathwayto decompress gastric and esophageal varices. In the im-mediate postoperative period, this may not provide anadequate outflow tract. All three patients who bled afterDSRS + SPD did so from gastric varices near the shortgastric vessels. The overall incidence of postshunt varicealbleeding has not been increased to date, and, while thepathophysiology may be different, we project that shortgastric hypertension should be no more of a temporaryclinical problem than renal vein hypertension.'6

In summary, these data show that addition ofcompletedissection of the splenic vein plus interruption of thesplenocolic ligament can preclude the development ofpancreatic siphon and result in better preservation ofpor-tal perfusion after selective shunt. This allows the follow-ing conclusions: (1) DSRS + SPD preserves postoperativeportal perfusion in the alcoholic and nonalcoholic cir-rhotics. (2) SPD maintains portal perfusion in alcoholics

al IIt

Vol. 204 * No. 4 SPLENOPANCREATIC DISCONNECTION 353(p < 0.05) better than the standard DSRS. (3) The he-modynamic and metabolic responses to DSRS + SPD aresimilar in the alcoholic and nonalcoholic population. (4)Hemodynamic stability and maintenance of the quanti-tative functions may be explained by better perfusion ofthe hepatocytes with portal blood and its hepatotrophicfactors. (5) Five-year survival ofalcoholics and even non-alcoholics may be improved after splenopancreatic dis-connection.

Acknowledgments

The authors acknowledge the assistance ofBeverly Noe, M.T.(ASCP),Bettye Hollins, Leslie Tilley, R.N., Joyce Oglesbee, R.N., Ruth Phillips,R.N., Janet H. Keen, R.N. Martha Del Sordo, and Joseph Jackson.

References1. Johnson GW, Rogers HW. A review of 15 years' experience in the

use of sclerotherapy in the control of acute hemorrhage fromesophageal varices. Br J Surg 1973; 60:797-800.

2. MacDougall BRD, Westaby D, Theodossi A, et al. Increased long-term survival in variceal haemorrhage using injection sclero-therapy: results of a controlled trial. Lancet 1982; 16 (January):124-127.

3. Terblanche J, Northover JMA, Bornman P, et al. A prospectivecontrolled trial of sclerotherapy in the long term management ofpatients after esophageal variceal bleeding. Surg Gynecol Obstet1979; 148:323-333.

4. Paquet K-J. Prophylactic endoscopic sclerosing treatment of theesophageal wall in varices: a prospective controlled randomizedtrial. Endoscopy 1982; 14:4-5.

5. Mitchell KJ, MacDougall BRD, Silk DBA, Williams R. A prospectivereappraisal of emergency endoscopy in patients with portal hy-pertension. Scand J Gastroenterol 1982; 17:965.

6. Novis BH, Duys P, Barbezat GO, et al. Fiberoptic endoscopy anduse ofthe Sengstaken tube in acute gastrointestinal haemorrhagein patients with portal hypertension and varices. Gut 1976; 17:258.

7. Johnson AG, Simms JM, Stoddard CJ. Is there a role for injectionsclerotherapy in the presence of active bleeding? In Westaby D,MacDougall BRD, Williams R, eds. Variceal Bleeding. London:Pitman, 1982; 159-164.

8. Conn HO. Ideal treatment of portal hypertension in 1985. In Ben-hamou J-P, Lebrec D, eds. Clinics in Gastroenterology: PortalHypertension. London:WB Saunders, 1985; 14:259-288.

9. Cello JP, Grendell JH, Crass RA, et al. Endoscopic sclerotherapyversus portacaval shunt in patients with severe cirrhosis and var-iceal hemorrhage. N Engl J Med 1984; 311:1589-1594.

10. Terblanche J, Yakoob HI, Bornman PC, et al. Acute bleeding varices:a five-year prospective evaluation of tamponade and sclerother-apy. Ann Surg 1981; 194:521-530.

11. Warren WD, Henderson JM, Millikan WJ Jr, et al. Distal splenorenalshunt versus endoscopic sclerotherapy for long term managementof variceal bleeding: preliminary report of a prospective, ran-domized trial. Ann Surg 1986; 203:454-462.

12. Warren WD, Rudman D, Millikan WJ Jr, et al. The metabolic basisof portasystemic encephalopathy and the effect of selective versusnonselective shunts. Ann Surg 1974; 180:572-579.

13. Galambos JT, Warren WD, Rudman D, et al. Selective and totalshunts in the treatment of bleeding varices: a randomized con-trolled trial. N Engl J Med 1976; 295:1089-1095.

DIscuSSION

DR. KIYOSHI INOKUCHI (Fukuoka, Japan): Many thanks for the priv-ilege of the floor. I am very pleased to have listened to Dr. Millikan's

14. Rikkers LF, Rudman D, Galambos JT, et al. A randomized, con-trolled trial of the distal splenorenal shunt. Ann Surg 1978; 188:271-282.

15. Millikan WJ Jr, Warren WD, Henderson JM, et al. The Emoryprospective randomized trial: selective versus nonselective shuntto control variceal bleeding: ten-year follow-up. Ann Surg 1985;201:712-722.

16. Warren WD, Millikan WJ Jr, Henderson JM, et al. Ten years portalhypertensive surgery at Emory: results and new perspectives. AnnSurg 1982; 195:530-542.

17. Henderson JM, Millikan WJ Jr, Wright-Bacon L, et al. Hemody-namic differences between alcoholic and nonalcoholic cirrhoticsfollowing distal splenorenal shunt: effect on survival? Ann Surg1983; 198:325-334.

18. Warren WD, Millikan WJ Jr, Henderson JM, et al. Selective varicealdecompression after splenectomy or splenic vein thrombosis witha note on splenopancreatic disconnection. Ann Surg 1984; 199:694-702.

19. Inokuchi K, Beppu K, Koyanagi N, et al. Exclusion of nonisolatedisolated splenic vein in distal splenorenal shunt for preventionof portal malcirculation. Ann Surg 1984; 200:711-717.

20. Warren WD, Millikan WJ Jr. Selective transsplenic decompressionprocedure: changes in technique after 300 cases. ContemporarySurgery 1981; 18:11-29.

21. Henderson JM, Millikan WJ Jr, Wright L, et al. Quantitative esti-mation of metabolic and hemodynamic hepatic function: theeffects of shunt surgery. Surgical Gastroenterology 1982; 1:77-85.

22. Tygstrup N. Determination of the hepatic elimination capacity (Lm)of galactose by single injection. Scand J Clin Lab Invest 1966;92(Suppl): 118-125.

23. Henderson JM, Kutner MH, Bain RP. First-order clearance ofplasma galactose: the effect ofliver disease. Gastroenterology 1982;83:1090-1096.

24. Henderson JM, Heymsfield SB, Horowitz J, Kutner MH. Measure-ment of liver and spleen volume by computed tomography. Ra-diology 1981; 141:525-527.

25. Feigenbaum H. Clinical applications of echocardiography. ProgCardiovasc Dis 1972; 14:531.

26. Nordlinger BM, Nordlinger DF, Fulenwider JT, et al. Angiographyin portal hypertension: clinical significance in surgery. Am J Surg1980; 139:132-141.

27. Harris V, Galoma G, Unger R. Radioimmunoassay procedure usingUngers' 04A antiserum for glucagon. In Jaffe B, Behrman H,eds. Methods of Humane Radioimmunoassay. New York:Aca-demic Press, 1979; 6433-6656.

28. Snedecor GW, Cochran WG. Statistical Methods, 7th ed. Ames,Iowa: Iowa State University Press, 1980.

29. Potts JR III, Henderson JM, Millikan WJ Jr, et al. Restoration ofportal venous perfusion and reversal ofencephalopathy by balloonocclusion of portal systemic shunt. Gastroenterology 1984; 87:208-212.

30. Millikan WJ Jr, Henderson JM, Potts JR III, et al. Liver functionand hemodynamic status one year into sclerotherapy for varicealbleeding. Gastroenterology 1984; 86:1184.

31. Zeppa R, Hensley GT, Levi JU, Livingstone AS. Factors influencingsurvival after distal splenorenal shunt. Ann Surg 1978; 187:510-542.

32. Faccioli AM, Maffei A, Gerundie GE. Preservation ofportal perfusionin distal splenorenal shunt (DSRS): the role of portosplenic dis-connection (personal communications).

33. Starzl TE, Porter KA, Francavella JA, et al. A hundred years of thehepatotrophic controversy. In Ciba Foundation Symposium.Hepatotrophic Factors. New York: Elsevier/Excerpta Medica/North Holland Publishing, 1978; 111-138.

fine presentation and to have looked at his beautiful operation ofspleno-pancreatic disconnection at Emory University Hospital the day beforeyesterday. I feel that you have overcome problems inherent to the distalsplenorenal shunt, such as loss of hepatic portal perfusion occurring after