Spleen stiffness: Toward a noninvasive portal sphygmomanometer?


  • Potential conflict of interest: Nothing to report.

Colecchia A, Montrone L, Scaioli E, Bacchi-Reggiani ML, Colli A, Casazza G, et al. Measurement of spleen stiffness to evaluate portal hypertension and the presence of esophageal varices in patients with HCV-related cirrhosis. Gastroenterology 2012;143:646-654. (Reprinted with permission.)


BACKGROUND AND AIMS: The hepatic vein pressure gradient (HVPG) is the standard used to determine the degree of portal hypertension (PH) and an important prognostic factor for patients with cirrhosis; HVPG values correlate with the presence of esophageal varices (EV). However, HVPG can only be accurately determined at specialized centers; noninvasive methods are needed to predict HVPG values and the presence of EV. We compared the diagnostic performance of spleen stiffness (SS) measurement by transient elastography with that of liver stiffness (LS) and of other recently proposed noninvasive tests. METHODS: We measured SS and LS in 100 consecutive patients with hepatitis C virus-induced cirrhosis. Patients were also assessed by FibroScan, HVPG, esophagogastroduodenoscopy, and liver biopsy. We also analyzed LS-spleen diameter to platelet ratio score and platelet count to spleen diameter. RESULTS: SS and LS were more accurate than other noninvasive parameters in identifying patients with EV and different degrees of PH. A linear model that included SS and LS accurately predicted HVPG values (R(2) = 0.85). The results were internally validated using bootstrap analysis. CONCLUSIONS: Measurement of SS can be used for noninvasive assessment and monitoring of PH and to detect EV in patients with hepatitis C virus-induced cirrhosis.


Patients with cirrhosis have a long period of compensated disease in which symptoms are minimal and the risk of death very low (median survival >12 years). In contrast, once the patient first decompensates, the probability of death is >50% within 2 years. 1 There are robust prognostic models to predict outcomes in patients with decompensated cirrhosis, but few tools to predict events in compensated patients. The increase in the portal pressure gradient (evaluated in clinical practice by the hepatic venous pressure gradient [HVPG]) over 10 mmHg is a major risk factor for the development of varices, decompensation, and hepatocellular carcinoma. Thus, this value has been used to define clinically significant portal hypertension (CSPH). 2 The presence of varices also reflects disease progression and worse prognosis in compensated patients. 1 In addition, the diagnosis of varices is relevant because there are effective treatments for the prevention of bleeding.

The standard technique to diagnose CSPH is to perform a hepatic vein catheterization, which is invasive, requires specific training and equipment, and is not available in many centers. The diagnosis of varices requires an upper gastrointestinal endoscopy, which is costly and can be uncomfortable. Consequently, it would be desirable to have noninvasive tools to assess the degree of portal hypertension (PH) and to predict varices, 2 and several have been proposed. Among them, liver stiffness measurement with transient elastography (TE; FibroScan) has shown value in predicting varices and correlates with HVPG, but is not accurate enough to predict HVPG values, especially in severe PH. 3, 4 This might be explained by the fact that liver stiffness would only reflect the contribution of increased hepatic resistance to PH, but not that of increased blood flow, more prominent in advanced PH. 4 Other noninvasive predictors take into account spleen diameter. These include the platelet/spleen ratio 5 and, most recently, the LSM–spleen diameter/platelet ratio score (LSPS) index. 6, 7 None have permeated clinical practice because of insufficient accuracy and lack of sufficient validation, and the recommendation to perform an upper endoscopy in all patients with cirrhosis still stands. 8

The study by Colecchia et al. 9 expands the array of available noninvasive markers of PH in cirrhosis with the evaluation of spleen stiffness (SS). The rationale is that PH leads to spleen congestion and organ congestion, by itself, is sufficient to increase organ stiffness. 10 In addition, recent experimental data showed that PH induces splenic fibrosis, 11 which would also increase SS. Therefore, whereas liver stiffness would only reflect the contribution of increased resistance, SS would be directly sensing the consequences of PH. Indeed, magnetic resonance imaging spleen elastography provided the initial proof of concept for this assumption, 12 and this was confirmed later with TE. 13

Colecchia et al. evaluated 113 patients with compensated cirrhosis resulting from hepatitis C. All had measurements of liver and spleen stiffness by TE, HVPG, and upper endoscopy within 1 week. Operators of one technique were blinded with respect to the other techniques. The researchers calculated also the platelet/spleen ratio and LSPS. SS was measured using the same TE probe used for liver stiffness. Ultrasonography (US) was used to localize the optimal point of measurement, which required a splenic parenchymal thickness of at least 4 cm. In 13 patients, the researchers failed to obtain reliable SS values, in 7 cases because of obesity (liver stiffness was also not reliable), in 3 because of the interposition of lung or colonic gas, and in 3 because of a anteroposterior spleen diameter of less than 4 cm. Mean SS was much higher than mean liver stiffness (56 versus 23 kPa). SS had excellent inter- and intraobserver reproducibility in a pilot study in 30 independent patients.

Sixty-five percent of the patients had CSPH. Patients with CSPH had higher SS than patients without CSPH (59 versus 39 kPa). There was a good correlation between SS and HVPG, better than for any of the other noninvasive markers. The best noninvasive prediction of HVPG was obtained with a combination of liver stiffness and SS values in a linear equation (HVPG = −4.44 + 0.241*LS + 0.226*SS). The R2 of the model was 0.82 after internal validation, which suggests a very good accuracy.

Fifty-three percent of the patients had varices. Patients with varices had also higher SS (56 versus 37 kPa in those without varices). There were no differences in SS between patients with different degrees of varices (grade I, II, and III: 54, 63, and 60 kPa, respectively).

To evaluate diagnostic accuracy, the researchers compared the AUROCs (area under receiver operating characteristic curves) of SS, liver stiffness, LSPS, and platelet/spleen ratio for predicting varices and CSPH. SS significantly outperformed LSPS and platelet/spleen ratio, but not liver stiffness, in the prediction of varices and CSPH. In addition, the researchers selected two cutoffs for each predictor, one to rule out and one to rule in the diagnosis. These cutoffs were chosen as those with highest positive likelihood ratio (rule in) and lowest negative likelihood ratio (rule out). An SS <41.3 kPa could reliably rule out varices, and an SS <40 kPa could reliably rule out CSPH. The corresponding cutoffs for liver stiffness to rule out varices and CSPH were 16.4 and 16 kPa, respectively.

In summary, the new findings in this study are that in patients with compensated HCV-related cirrhosis (1) SS correlates with HVPG, (2) a model combining SS and liver stiffness can predict, with relatively high precision, HVPG values, and (3) SS is better than other noninvasive models to predict varices and CSPH, but does not significantly outperform liver stiffness.

This study undoubtedly provides important pieces of information, but leaves some doubts about the potential of SS. First, the technical success of SS measurement with the FibroScan apparatus was high (88%). In our limited and uncontrolled experience trying to measure SS with FibroScan, we had a much higher number of technical failures, mainly resulting from the inability to find a spot with at least 4 cm of splenic thickness. One possible explanation is that the patients in the present series were somewhat selected in the sense that they had more-advanced disease and thicker spleens. Indeed, the prevalence of varices (53%), much higher than in historical and recent 14 series of patients with compensated cirrhosis, hints that this could be the case. SS applicability in very early cirrhosis, and in patients with smaller spleens, such as those with alcoholic cirrhosis, 15 remains to be investigated. In addition, the need for US guidance is a limitation because it poses a challenge to a major advantage of elastography: its simplicity. In this regard, a very recent report suggests that the measurement of SS with a different sonoelastography method (acoustic radiation force impulse imaging), which allows to select the region of interest under real-time US control, achieves higher success rates than TE. 16 This finding still requires validation.

Second, this study fails to provide clear-cut evidence that SS would be better than (or add to) liver stiffness in the screening for varices or CSPH. Although the AUROC curves were better with SS than with liver stiffness (though nonsignificantly), this does not answer which one would be better as a screening tool. The AUROC for detecting varices with SS represents the probability that given any two patients, one with varices and one without varices, SS would be higher in the one with varices. However, this is seldom the question that would guide decision making, where the point is to predict the risk of varices in a given patient, not to discriminate between two patients. In this regard, it would have been desirable to know the number of patients correctly classified by liver stiffness and SS and to provide estimators of the improvement in classification achieved with SS. 17 In addition, the combination of liver stiffness and SS predicts HVPG values better than any of the two alone. This could suggest that a model combining liver stiffness and SS could be also better in predicting varices and CSPH, as compared with each predictor individually, but this was not evaluated.

Finally, as discussed by the researchers, these results would require validation in independent settings, ideally in patients with different etiologies of cirrhosis. It would be interesting to assess, in new longitudinal studies, whether SS is sensitive to disease progression and regression and whether it can reflect the response to treatments to decrease portal pressure, such as beta-blockers. These new studies should also confirm or refute whether the combination of liver and spleen stiffness could stand for the long-awaited noninvasive portal sphygmomanometer.