Hypothermic Oxygenated Perfusion Improves Extended Criteria Donor Liver Graft Function and Reduces Duration of Hospitalization Without Extra Cost: The PERPHO Study

Liver transplantation (LT) is the best treatment for end-stage liver disease and primary liver malignancies. Because of the growing gap between the number of candidates and transplant recipients, transplantation teams have been forced to increase the donor pool by using more and more grafts from so-called extended criteria donors (ECDs).

Although they may currently represent more than 50% of the grafts used,⁽¹⁾ there is still no consensual definition of an ECD graft. However, they are mostly represented by elderly donors, donation after circulatory death (DCD) donors, or fatty liver grafts and are well known to be more vulnerable to ischemia/reperfusion injuries (IRIs)⁽²⁾ created during conventional static cold storage. These grafts are therefore associated with higher rates of graft dysfunction.^(3-6)

As a result, there has been increasing interest in perfusion machines to improve the quality of conservation and reduce the consequences of IRI. Among the different techniques of perfusion, hypothermic oxygenated perfusion (HOPE) is simple and has already proved its efficacy in preserving DCD grafts by reducing biliary complications and improving graft and patient survival.⁽⁷⁾ However, few studies have evaluated the interest of HOPE in preserving ECD grafts procured from donation after brain death (DBD) donors or the cost from a hospital’s perspective because the procedure is not covered by health insurance in most countries.

We hypothesized that HOPE could benefit LTs performed with ECD grafts and implemented a prospective trial, the Oxygenated Hypothermic Perfusion in Preservation of Hepatic Grafts From Expanded Criteria Donors (PERPHO) Study, to confirm it.

Patients and Methods

Results

Recipient and Graft Characteristics

Between February and November 2018, 97 LTs were performed in our institution. An ECD graft was used in 54 (56%) cases. After taking into consideration exclusion criteria, 43 LTs could have been potentially included in the PERPHO Study; among them, 25 patients were finally included and constituted the HOPE group. The remaining 18 LTs were already included in another study⁽¹⁴⁾ or were not included because of operator availability for the machine perfusion (Fig. 1).

Patients were men in 80% (n = 20) of the cases, with a median age of 63 (range, 43-69) years and a median MELD score before LT of 18.3 (range, 7-37). Of the patients, 72% (n = 18) were at home prior to LT, and 28% (n = 7) were hospitalized, including 16% (n = 4) in the ICU with intubation. Of the donors, 54% (n = 14) were male, with a median age of 70 (range, 45-87) years, and 19 (76%) were >65 years of age (Table 1).

TABLE 1. Characteristics of Recipients and Donors in the HOPE Group and Postoperative Outcomes

	recipient characteristics	donor characteristics				intraoperative			postoperative outcome
patient number	Age, Years	Sex	CTP Grade	MELD Score at LT	Indication for LT	Age, Years	BMI	Cardiac Arrest	Steatosis	Perfusion Duration (Minutes)	CIT (Minutes)	PRS	EAD	PNF	Clavien-Dindo	Hospital Stay (Days)	3-Month Graft Survival	3-Month Patient Survival	Cause of Graft Loss and/or Death
1	64	Male	5	7	HCC + ALD	80	23.8	No	5	93	642	Yes	Yes	No	1	8	Yes	Yes	–
2	63	Female	9	15	ALD	68	43	No	30	109	536	No	No	No	3	24	Yes	Yes	–
3	63	Male	7	9	HCC + ALD	83	28.2	No	0	75	425	No	No	No	2	12	Yes	Yes	–
4	61	Male	6	10	HCC + ALD	69	25.2	No	15	125	491	No	Yes	Yes	4	97	No	Yes	HV thrombosis
5	62	Male	12	17	HCC + ALD	47	25	Yes	0	115	630	No	No	No	2	21	Yes	Yes	–
6	64	Male	6	22	HCC + ALD	83	29	No	0	105	616	Yes	No	No	1	10	Yes	Yes	–
7	58	Male	13	34	ALD	87	18	No	10	95	511	Yes	No	No	2	13	Yes	Yes	–
8	66	Female	12	18	ALD	75	26.2	No	10	97	414	Yes	No	No	3	13	Yes	Yes	–
9	49	Male	15	34	ALD	76	22.6	No	0	90	526	Yes	Yes	No	2	33	Yes	Yes	–
10	43	Male	12	25	ALD	64	32.4	No	5	117	554	No	No	No	2	13	Yes	Yes	–
11	50	Male	5	16	HCC + ALD	53	21.6	Yes	5	226	692	No	No	No	2	19	Yes	Yes	–
12	63	Female	11	37	ALD	68	25.3	No	5	130	379	Yes	No	No	2	15	Yes	Yes	–
13	63	Male	9	12	HCC + ALD	70	29.4	No	0	252	816	Yes	Yes	No	3	19	Yes	Yes	–
14	57	Male	8	19	ALD + HCV	84	29	No	15	222	547	Yes	Yes	No	2	24	Yes	Yes	–
15	62	Male	12	21	ALD	70	20.7	No	0	178	525	Yes	No	Yes	3	23	No	Yes	HA thrombosis
16	66	Male	5	12	HCC + ALD	70	27.3	No	0	130	615	Yes	No	No	2	13	Yes	Yes	–
17	57	Male	12	22	HCC + ALD	68	23.4	Yes	0	116	512	No	No	No	2	15	Yes	Yes	–
18	68	Male	9	17	HCC + ALD	82	23.7	No	5	195	530	No	No	No	2	12	Yes	Yes	–
19	59	Male	10	22	Other	71	22.1	No	0	124	444	No	No	No	2	10	No	No	Multivisceral failure
20	65	Male	5	18	HCC + ALD	51	25.1	Yes	NA	107	495	No	Yes	No	2	20	Yes	Yes	–
21	62	Male	6	14	HCC + ALD	82	20.3	No	0	144	505	Yes	No	No	2	8	Yes	Yes	–
22	56	Female	12	32	NASH	79	24.5	No	0	153	665	No	No	No	2	15	Yes	Yes	–
23	66	Male	11	18	ALD	78	22.9	Yes	10	100	519	Yes	No	No	3	13	Yes	Yes	–
24	69	Female	11	20	ALD	48	25.4	No	0	124	458	No	No	No	2	17	Yes	Yes	–
25	66	Male	5	8	HCC + ALD	45	40.9	No	NA	90	445	Yes	Yes	No	2	11	Yes	Yes	–

Predictive Value of Liver Enzyme Levels Measured on Perfusate at the End of Perfusion

Liver enzyme dosage on the perfusion solution was analyzed for 20 patients in the HOPE group (dosage not performed or failed in 3 cases and excluded in 2 cases [patients 4 and 15] because of the presence of vascular complications, which could impact postoperative biological parameters). The median AST level in the perfusion solution was 248.5 (range, 48-2076) U/L, and the median ALT level was 285.5 (range, 38-1397) U/L.

Among the 20 patients, 4 patients presented with EAD and had a significantly higher AST level (980.5 U/L versus 170 U/L; P = 0.01) and ALT level (1073.5 U/L versus 143.5 U/L; P = 0.01) in the perfusion solution compared with the patients who did not present EAD. The ROC curve analysis (Fig. 2) revealed that AST and ALT levels in perfusate were excellent predictors of EAD occurrence with areas under the curve (AUCs), respectively, of 0.92 and 0.91. The best predictive cutoff was 800 U/L for AST and ALT. As a consequence, postoperative blood liver enzymes were significantly increased in the early PODs when liver enzyme levels were increased in the perfusate (Fig. 3).

Postoperative Parameters

The median AST level was significantly lower in the HOPE group on POD 0 (724 U/L versus 1284 U/L; P = 0.046), as were the ALT (392 U/L versus 720 U/L; P = 0.01), lactate (2.2 versus 2.7; P = 0.01), and creatinine level (73 versus 89 µmol/L; P = 0.01). On POD 1, the lactate level was still lower in the HOPE group (1.5 versus 2; P = 0.03), whereas the difference was no more significant regarding the AST (599 U/L versus 870 U/L; P = 0.09) and ALT (454 U/L versus 668 U/L; P = 0.06; Fig. 4).

The acute kidney injury incidence (calculated at 48 hours after LT) was no different between groups (44% [n = 11] versus 28.9% [n = 20]; P = 0.17). As for cholestasis enzymes, the gamma-glutamyltransferase (GGT) level was significantly increased in the HOPE group from POD 5 (253 versus 191; P = 0.04) as in the alkaline phosphatase (ALP) level on POD 7 (186 versus 129; P = 0.024), whereas the bilirubin levels were not different.

Median ICU stay was significantly reduced in the HOPE group (3 versus 5 days; P = 0.01), as was hospital stay (15 versus 20 days; P = 0.01). Of the patients, 6 (24%) in the HOPE group and 31 (44.9%) in the control group had severe complications (ie, Clavien-Dindo score ≥3), without significant difference (P = 0.07).

The occurrence of biliary complications during the first 12 months was similar between groups (8% [n = 2] versus 11.6% [n = 8]; P = 1). In the HOPE group, 2 patients presented with anastomotic complications requiring surgical revision for bile leakage in 1 case and endoscopic treatment for anastomotic stricture in the other. In the control group, 7 patients presented with anastomotic complications represented by 4 bile leakages (requiring surgery in 2 cases and medical treatment only in 2 cases) and 3 anastomotic strictures treated by surgery in 1 case and endoscopic treatment in 2 cases. One patient in the control group presented with biliary necrosis requiring retransplantation. No patient presented with nonanastomotic stricture.

At 12 months, graft survival was similar between the 2 groups (88% in the HOPE group versus 89.5% in the control group; P = 1.00), as was patient survival (91% in the HOPE group versus 91.3% in the control group; P = 1.00).

Discussion

LT is the best treatment for end-stage liver disease, but it is limited by the scarcity of grafts. Expanding the donor pool with DCD or ECD grafts appears to be the most effective solution despite their higher risk of dysfunction. To decrease this risk and improve outcome, machine perfusion has been evaluated and shown promising results.^{(15, 16)}

The PERPHO Study is the first prospective trial evaluating the interest of HOPE in the preservation of ECD grafts procured from DBD donors. The study was not designed to test an increase of the total conservation time or to increase the number of LTs performed by reconditioning discarded grafts.

We found a significant reduction in AST and ALT levels in the early postoperative period, attesting to a reduction in hepatocyte damage, as well as a significant decrease in lactate levels, reflecting faster graft function recovery. However, we did not note a significant decrease in EAD incidence (28% in the HOPE group versus 42% in the control group), which was probably attributed to a lack of power. Interestingly, liver enzyme levels in the perfusion liquid, measured at the end of the HOPE procedure, were highly predictive of EAD incidence and correlated to postoperative serum liver enzymes.

In addition, we observed that the serum level of ALP and GGT (as well as the bilirubin level) started to increase from POD 2 in both groups. However, this increase was higher in the HOPE group and became significant at POD 5 for GGT and POD 7 for ALP before spontaneously resolving. We believe this phenomenon may represent the liver regeneration, which was described as positively correlated with the serum level of a biliary marker⁽¹⁷⁾ instead of a marker of biliary damage.

We also observed a significant reduction in ICU (3 versus 5 days; P = 0.01) and total hospital stays (15 versus 20 days; P = 0.01). This could be explained by better graft function recovery as well as the reduced incidence of severe complications, which was close to being significant (P = 0.07). Consequently, from a hospital’s perspective, the additional cost of the machine perfusion procedure (estimated at €5298 [US $6286] per patient) was compensated by a reduction of hospitalization length, resulting in a nonsignificant difference between cost and revenue between the 2 groups. This analysis was made from the hospital’s perspective (and not the health care system’s perspective) because the decision to invest or not into the machine perfusion program, with the extra costs, which is the main obstacle to its implementation, is up to the hospitals with no participation of the national health care systems in most countries.

Moreover, even if the aim of the study was not to increase the number of LTs performed, we believe that 1 procedure was only allowed by the HOPE procedure. Indeed, in the case of patient 13, an unpredictable weather event led to the graft arriving with more than 10 hours of cold ischemia time and without immediate access to an operating room (because of another urgent procedure). In this case, the graft would have probably been considered unfit for transplantation without the potential benefit of HOPE because we would not have risked using an ECD graft with such prolonged ischemia. Instead, we decided to put the liver on machine perfusion and wait until another operating room was available. As a consequence, we did not find a suitable control for this case, highlighting its exceptional nature. In the same way, when the hepatectomy is complicated, we believe that HOPE makes it possible to partially mitigate the stress and the pressure of cold ischemia.

Another change of paradigm was represented by recipient-graft matching. Indeed, ECD grafts are usually allocated to “good” recipients (mostly hepatocellular carcinoma [HCC] patients with compensated cirrhosis), whereas most “high-MELD” recipients receive non-ECD grafts. However, in our study, 3 patients who received a transplant had a MELD score >30 with grafts from donors >75 years of age. Because the postoperative outcomes were uneventful, we believe that HOPE makes it possible to safely allocate ECD grafts to “high-MELD” recipients.

Similar to kidney transplantation, liver perfusion started in the beginning of this decade with the first clinical series in 2010 reported by Guarrera et al.⁽¹⁵⁾ In their pilot study, the authors reported the use of hypothermic machine perfusion without oxygenation for conservation of DBD “standard” donors and found an improvement in biological markers and shorter hospital stays in the machine group. Thereafter, the same team reported their experience with 31 liver grafts considered as “orphan”⁽¹⁸⁾ and reported higher 1-year patient survival and significantly shorter hospital stays as well as fewer biliary complications and acute kidney injuries. However, Schlegel et al.^{(19, 20)} showed that HOPE recharged depleted cellular energy stores (ie, ATP level), restored the mitochondrial redox state by reversible suppression of oxidative metabolism, and subsequently decreased the production of oxygen radical species, resulting in less IRI and improved liver function.⁽²¹⁾ Thereafter, machine perfusion proved its utility in LT with DCD grafts by improving graft function as well as survival compared with conventional static conservation.^{(7, 22)} Recently, Nasrala et al.,⁽²³⁾ in a prospective randomized trial using normothermic perfusion for preservation of all sorts of grafts, reported a significant decrease in postoperative liver enzyme levels as well as in the incidence of EAD and the rate of discarded grafts. However, despite the fact that normothermic machine perfusion is supposed to be more efficient than HOPE (but also more complex), they failed to show any clinical improvement, as the duration of hospital stays as well as the graft and patient survival were no different. One could therefore think that machine perfusion is not required for all LTs, especially those with “non-ECD grafts.” Recently, Patrono et al.⁽²⁴⁾ reported their results of 25 patients who received ECD grafts using dual HOPE. They noted a significant reduction in postoperative AST and ALT blood levels, resulting in a decrease in EAD incidence and a significant reduction in stages 2 and 3 acute kidney injuries. However, they did not report a significant difference regarding hospital stay and survival, which could be partly explained by the significantly higher duration of preservation in the perfusion group (499 versus 371 minutes; P < 0.001).

Our results are in line with studies relating the benefits of the HOPE procedure, as we noted a reduction in liver enzyme levels^{(23, 24)} and better graft function recovery. However, we found that HOPE significantly reduced hospital stays, resulting in equivalent hospitalization costs. Therefore, as with transplantations with DCD grafts,^{(7, 22)} we believe that use of machine perfusion is justified and economically acceptable for LT with DBD-ECD grafts.

However, our results must be interpreted with caution. First, because this was a pilot study, we had a limited number of included patients (n = 25). Moreover, we only found 69 patients for the control group (as a result of caliper restriction) instead of 75, which decreased the power of our study. However, because the unmatched patients of the HOPE group (which could be considered as having a poorer prognosis) were kept in the analysis, we believe that our results may underestimate the potential benefit of HOPE. Second, our control group was not prospective, which could induce a methodological bias in both the clinical and cost analysis. However, we voluntarily decided to not create a prospective control group because we believe that a prospective randomized trial with only 25 patients in both the HOPE and control groups has a high risk of leading to unbalanced randomization and then to noncomparable groups. Third, our definition of ECD graft could be questionable and may not reflect the real quality of the graft. However, there is still no consensual definition of an ECD graft, and we previously reported those criteria, which are now used in prospective trials in our country.^{(14, 25)} In any case, our promising results must be confirmed in a prospective randomized trial, which is currently ongoing.⁽²⁵⁾ Fourth, the wide range of the HOPE duration in our study may impact the liver enzyme level in the perfusate. However, because the plateau level is quickly reached and stays mostly stable until the end of perfusion⁽¹⁸⁾, we believe that the HOPE procedure duration does not represent a major bias. Finally, our cost analysis finding may not be directly applicable to other health care systems because of the specificities of the French hospital funding. However, because our estimation of the extra cost of the HOPE procedure will be the same whatever the country or the health care system, we believe that our results will provide informative data for further studies.

In conclusion, we believe that HOPE is a promising method to improve the preservation of DBD-ECD grafts because it may provide better graft function. In our study, the additional cost of the procedure was compensated by better outcomes. These promising results must be confirmed in a prospective, multicentric, randomized trial.