Failure of splenectomy to ameliorate portal hypertension in rnyeloproliferative disorders

The correct treatment of portal hypertension associated with myeloproliferative disorders remains uncertain. Splenectomy has been advocated by some to eliminate the forward flow component of the portal hypertension and thus reduce portal pressure. The authors describe three recent cases of myeloproliferative disorder in whom splenectomy failed to achieve any significant amelioration of portal hypertension, with in-depth hemodynamic studies in one patient. Based on these experiences, the authors suggest that splenectomy is not the optimum treatment of the portal hypertension 
associated with myeloproliferative disorders.

occasionally arc associated with portal hypertension (1 -9).Indeed, consequences of the portal hypertension, such as esophageal variceal blee<ling and asc ires, often become the dominant clinical featu res in these patients.The pathogenesis of portal hypertension in these cases is unknown.Previous studies had suggested that splenomegaly and consequent increased splenic bloodflow augmented portal pressure ( 10,11), which led some to propound splenectomy as a treatment for portal hypertension m various hematological diseases.We have treated three patients with myeloproliferative disorder and portal hypertension in whom splenectomy had no effect on the course of portal hypertension.Detailed hemodynamic studies were done on one patient.CASE PRESENTATIONS Case 1: A 60-year-old Caucasian male was admitted to hospi tal with hematochezia.He had been generally well until one year before admission, when ascites and spleno megaly were no ted.Past history included adult-onset diabetes and recurrent duodenal ulcers which responded to vagotomy anJ pyloroplasty.
A computed tomographic scan done fo ur months before admission showed an enlarged portal vein, splenic and paragastric varices, and cho lclithiasis.A tentative diagnosis of 'cryptogenic cirrhosis' was maJe, bur liver biopsy was not done because of the ascites.
Physical examination revealed ascites, peripheral edema and massive splenomegaly.The remainder of the examination was unremarkable; stigmata of chronic li ver disease were absent and the li ver could not be palpated.
His hemoglobin was 11 2 g/L, white blood cell count was 15.0xl0 9 {L and platelet count was 687xl0 9 /L.The diagnosis of myelofibrosis was made from examination of the blood smear and bone marrow biopsy.Gascroscopy done shortly after admission confirmed a bleeding origin from esophageal varices.
One month after admission, the patient had a splenectomy in an effort to ameliorate the portal hypertension.At laparotomy, the liver was noted to be faintly nodular and firm but did not demonstrate any gross evidence of cirrhosis.A needle biopsy of the liver revealed nodular regenerative hyperplasia.The histology of the spleen was typica l of myelofibrosis.The postoperative course was complicated by marked thrombocycosis which was treated by leukapheresis anJ h yJroxyurea.
On the sixth postoperative day, the patient haJ a further variceal hemorrhage, and demonstrated worsening of the asc ites and peripheral edema.The possibilities of hepatic or portal vein thrombosis as a consequence of thrombocytosis were entertai ned, but Doppler ultrasound scanner was not available.
Over the next four weeks, despite aggressive medical treatment, the patient <leteriorate<l .He continued to have recurrent variceal bleeding, increasing ascites and fluid retention, and <lied on the 34th postoperative day.Permission for autopsy was refused.Case 2: A 67-year-old Caucasian male was admitted with hematemesis.Three years previously, the diagnosis of 98 myelofibrosis had been made after he presented with uncontrolled bleeding following dental extraction , and he had thrombocytosis and leukocytosis (bone marrow biopsy confinned the diagnosis).Past history also included hypothyroidism and gout.
Physical examination showed pallor of the skin, splenomegaly, ascites and evidence of hypothyroidism.The remainder of the examination was unremarkable.
His hemoglobin was 70 g/L, white blood cell count was 54.0x10 9 /L and platelet count was 82x10 9 [L.Technetium liver scan and ultrasound of the li ver were normal aside from splenomegaly and ascites.Gastroscopy revealed large esophageal varices and angiography showed a grossly dilated portal venous system and splenomegaly with dilated splenic artery.No thromboses were seen.
After a<lmission, hematemesis ceased spontaneously and the ascites resolved with diuretic therapy.Percutaneous liver biopsy demonstrated changes consistent with early nodular regenerative hyperp lasia.
In view of the portal hypertension and thrombocytopenia, a splenectomy and a proximal splenorenal shunt were done.Postoperatively, the patient made a good recovery and was Jischarged.He was readmitted four times in the following two years with further episodes of variceal bleeding, which stopped either spontaneously or with endoscopic sclerotherapy.Patency of the splenorenal shunt could not be determined noninvasively, and he refused further angiography.
O n the patient's final admission, variceal bleeding did not stop spontaneously.He refused sclerotherapy, but consented to esophageal transection, which stopped the bleeding.He was initially well postoperatively, but then developed fever and suspected intra-ab<lo minal sepsis.The patient died with multisystem organ failure, and consent for autopsy was refused.Case 3: A 78-year-old Caucasian female was admitted fo r investigation of hypersplenism.She complained of pruritus, weakness, night sweats and fever.Other than a history of remote self-limited hepatitis with no obvious sequelae, there was no history of liver disease.
Physical examination revealed small, firm lymph nodes in the left anterior cervical chain and massive splenomegaly with an overlying bruit.The skin was excoriated.T he reminder of the examination was normal.
Her hemoglobin was 82 ~. white blood cell count was 2.2xl0 9 /L and platelet count was 240xl0 9 /L.She was iron deficient.A bone marrow biopsy showed hypercellularity and increased reticulin but no sign of malignancy.The patient was discharged on oral iron therapy with a diagnosis of myelodysplastic syndrome.
Three months later, she was readmitted with hematemesis and melena, and was found to have large esophageal varices on endoscopy.The bleeding stopped with sclerotherapy.Because of the disappointing experience with the rwo previous cases, the authors felt that complete hemodynamic assessment would be useful before considering any further interventions.Accordingly, the patient was readmitted one month later for a transjugu lar liver biopsy and hepatic hemodynamic stud ies.The hepatic venous pressure gradient was elevated at 12 mmHg ( wedged and free hepatic venous pressures were 20 and 8 mmHg, respectively).The cardiac output was slightly elevated at 6.6 L/min.A splenoportogram revealed patency of the splenic and portal veins, and extensive portosystemic collateral formation.The splenic pulp pressure was 24 mmHg.T he biopsy specimen was fragmented but consistent with nodular regenerative hyperplasia.
ln view of the cytopenia, thought to be due to hypersplenism, and the possibility of ameliorating the portal hypertension, the patient was booked fo r splenectomy.O ne month after the hemodynamic studies, a splenecromy and a proximal splenorenal shunt were done.Prior to the shunting procedure, the intraoperative portal pressure was 21 mml-lg; this fell to 11 mmHg after the anastomosis was created.Histology of the spleen revealed a large cleavedcell follicular lymphoma.An operative wedge biopsy of the liver confirmed the presence of nodular regenerative hyperplasia (Figure 1).
Postoperatively, a computed tomographic scan of the abdomen revealed portal vein thrombosis.Endoscopy showed small esophageal varices but when repeated three days later, the varices had regrown to their preoperative size.Therefore, three weeks after the initial operation, a side-to-side portocaval shunt bypassing the portal vein thrombosis was performed.The portal pressure before the shunt was established was 22 mmHg; after the anastomosis it fell to 8 mmHg.The postoperative course was uneventful and two weeks after the second operation, varices were absent on repeat e ndoscopy.Doppler ultrasound showed flow in the portocaval shunt although the clot was still present.
After a course of intravenous heparin, the patient was discharged on oral warfarin.Repeat endoscopy one year later revealed no esophageal varices.Her current hemogram, including white blood cell and platelet counts, is within normal limits.Because her symptoms are quite indolent, she is not undergoing chemotherapy for her lymphoma, and has demonstrated no e vidence of liver failure, with no further bleeding during the two-year follow-up.

DISCUSSION
The best treatment fo r portal hypertension in patients with myeloproliferative disorder remains unclear, but splenectomy has been advocated by some, based on anecdotal reports and theoretical reasons why it should be effective.Based on our experience with the three presented cases and review of the theoretical aspects, we disagree with this notion.
The pathogenic mechanisms inducing portal hypertension in myeloproliferative disorders are still unknown.Earlier studies indicated that the massive splenomegaly in myeloproliferative disorders is associated with increased splenic bloodflow: up to 1.5 L/min (10,11).
Because portal pressure (P) is the product of portal venous flow (Q) and intrahepacic portal resistance (R), according to the fo rmula P=QxR, it Splenectomy In portal hypertension Figure 1) Liver biopsy from case 3. The siriu.s red stain for connecrive tissue shcxus J>ortions of several parenchymal nodules separated by compressed lit•er cells but noc by fibrous septa, 1yp1cal of nodular regenerative hyperf>lasra would be expecte<l thaL increasing flow into the portal venous syste m would also increase portal pressure.This is known as the 'forward flow' mechanism (1 , 11), while the 'backward flow' mecha nism comprises facto rs that increase resistance.le is generally accepted that most forms of portal hypertension in humans are due to a combination of the two factors.S ince it appears that he patic sinusoidal circulation is inte rmittent, with only 20 to 25% of sinusoids being perfused at any instant, hepatic vascular reserve capacity is considerable (1 2).This explains experimental data that show that huge increments in flow, even doubling or trebling flows, result in only modest increases in portal pressure ( 13, 14 ).Thus it appears that only under exceptional circumstances can forward flow -by itself -cause significant portal hypertension.lnterestingly, the forwa rd flow idea led some workers t0 try sple nectomy in portal hypertension due to cirrh0sis three to four decades ago.H owever, the comple te lack of efficacy led to abandonment of this procedure.
Myeloproliferative disorders are known to be associated with increased clotting risk, an<l operative intervention in particular ca rries high rbk of thromboses in the portal ven0us system.In two of our cases there was circ umstantial evidence, and in t he third, proof of clots in th e portal system.
Recent studies have shown that th e liver is abnormal, even when not infiltrated by tumour cells, in myeloproliferative disorde rs.The sinusoids may show subtle morphological c hanges predisposing to portal hypertension, such as perisi nusoidal fibrosis or collagenization of the Disse space (15).
Wanless and colleagues ( 16) fou n<l th at hepatic microvascular thrombi arc much more common than previously thought, with 40% of au topsied cases of myeloproliferative syndrome showing these lesions.Such abnormalities likely lead to presinusoidal and sinusoidal portal hypertension and nodu larity, as typified by nodular regenerative hyperplasia.T his lesion was present m all three of our cases.S ince nodular regenerative hyperplasia has been well <lefi ned only within the past decade (l 7), one might speculate that some of the previous reported cases of ponal hypertension in hematological disorders may have been unrecognized examples of this condition.Nodu lar regenerative hyperplasia causes portal hypertension predominantly by backward flow mechanisms (sinusoidal compression by nodules ::ind vascular distortion), with a lesser contribution by forward flow (splenomegaly with increased splenic bloo<lflow) ( 18).
Finally, there appears to be an inverse inter-relationship between splenic anJ superior mesenteric venous flow such that overall portal venous flow tends to remain constant.Studies by Moriyasu and coll eagues in patients with portal hypertension, using a duplex Doppler system, show that splenectomy or transient occlusion of the splenic artery leads to augmentation of superior mesenteric venous flow, and a lower than expected decrease in portal pressure ( 19,20).T herefore, the overall impact of decreases in splenic bloodflow to the forward flow component of portal pressure will be blunted.
The outcome of splenectomy in our patients is at odds with several previous reports of successful eradication of varices after splenectomy in myeloproliferative disorders (5)(6)(7)(8).However, portosystemic collaterals are well known to arise in sites of intrabdominal surgery, and one wonders to what ex-