Why do we use the model for end-stage liver disease (MELD) score?: Part 2

In Part 1 of this series, we reviewed what came BEFORE the MELD score—Child-Turcotte-Pugh classification. We additionally discussed the history behind MELD, what is included in its calculation and why, and how it is used in transplant listing.

Is everyone excited for Part 2??

What are the exceptions to MELD allocation?

As discussed in Part 1, we know that the MELD score accurately predicts short-term three-month mortality in greater than 80% of waitlisted candidates. However, there are certain disease processes where mortality is actually higher than what would be captured by the MELD. How can we best serve this particular population of patients?



Hepatocellular carcinoma (HCC)
Hepatopulmonary syndrome
Portopulmonary hypertension
Familial amyloid polyneuropathy
Primary hyperoxaluria*
Cystic fibrosis
Hilar/perihilar cholangiocarcinoma
Hepatic artery thrombosis*

If patients have any of these conditions (and there are specific criteria for each one), they are eligible for an upgrade—otherwise known as an exception—in the MELD score at which they are listed.

But what does an “upgrade” or an “exception” mean?

A MELD upgrade is defined as the median MELD score at transplant (MMaT) within a 250 nautical mile radius of the transplant minus 3. For example:

  • Patient A has a MELD score of 15 based on her labs. However, she has HCC that falls within the criteria (reviewed below) necessary to achieve exception points. The median MELD at transplant for her transplant center is 31. Therefore, her MELD upgrade will be (31-3)=28. She will be listed with a MELD of 28 rather than 15.

There are two conditions (* in table above)—hepatic artery thrombosis (HAT) and primary hyperoxaluria—whereby patients are not listed at MMaT-3.

Patients with HAT that occurs within 7 days of transplantation (along with AST≥3,000 U/L and at least one of INR ≥2.5, arterial pH≤7.30, venous pH≤7.25, or lactate≥4 mmol/L) get listed as Status 1A. If HAT occurs within 14 days after transplantation (even if it does not meet the other requirements), patients automatically get listed at a MELD of 40.

For primary hyperoxaluria, patients are listed at MMaT. Three points are not subtracted.

The differences in listing for these two conditions is because of the very high mortality that they are associated with.

Most common reason for MELD Exception Points: HCC

As described in the example above with Patient A, patients with HCC often do not demonstrate the degree of hepatic dysfunction sufficient to give them a MELD score high enough for transplant. As we know, listed patients with low MELD scores often wait seemingly indefinitely for transplantation. This long waiting period in patients with HCC would lead to tumor growth and potentially metastases, thereby increasing mortality.

On the other hand, transplanting a patient with advanced HCC would not necessarily make sense given post-transplant survival may not be better than survival without a transplant. A person with a very large tumor that has invaded vessels or has spread with distant metastases is unlikely to benefit from transplantation. Thus, the development of Milan’s criteria for HCC exception points!

In 1996, Dr. Vincenzo Mazzaferro published a landmark study examining patients transplanted with HCC with the following restrictions: single lesion ≤5cm OR up to three separate lesions, none larger than 3cm; no evidence of gross vascular invasion; no regional nodal or distant metastases.

He found that the survival rate 4-years post-transplant was similar to the survival rate for patients undergoing transplantation without HCC. These criteria became known as MILAN CRITERIA.
Figure 1. Overall Survival (Panel A) and Recurrence-free Survival (Panel B) after Liver Transplantation in 48 Patients with Small Hepatocellular Carcinomas and Cirrhosis.
This is where we got the number of 75% for post-transplant and recurrence free survival
Figure 2. Effect of Anticancer Treatment before Transplantation on Overall Survival (Panel A) and Recurrence-free Survival (Panel B) in 48 Patients with Cirrhosis and Hepatocellular Carcinomas.
The dotted line demonstrates worse survival when criteria for tumor size was not met. This is how Milan criteria became accepted!

I understand the survival benefit, but why were the particular sizes in Milan criteria chosen?

In the original study’s methodology, Mazzaferro and his colleagues describe, “For patients with a single hepatocellular carcinoma to be eligible for the study, the tumor could not exceed 5 cm in diameter. In patients with multiple tumors, there could be no more than three tumors, none exceeding 3 cm in diameter. Patients in whom tumor invasion of blood vessels or lymph nodes was evident or suspected preoperatively were excluded.” This does not explain, however, how these sizes were chosen.

To find out the answer to my question, I decided to e-mail Dr. Mazzaferro and ask!

Within one hour of my e-mail, I received an insightful response. The highlights are below:

During my training years in Pittsburgh with doctor, Thomas Starzl, I reviewed (and never published) the original experience of that Center with liver transplant (LT) in HCC. It was a difficult data set to look into but at the end, it turned out that the earlier was the tumor stage at the time of transplant (incidental HCC was the definition of early HCC at that time), the better was the observed patient outcome.

I spent long hours in trying to find a “cutoff” value of tumor burden able to predict adverse post-LT but I could not reach a clear conclusion. Since transplant data did not suffice, I also looked back into the result of liver resection of HCC, both in my original Hospital in Italy and into the literature…A combination of common sense and observation of natural history of HCC suggested me to combine single AND multiple lesions of this tumor to predict its prognosis.

The cutoffs of Milan criteria were decided looking again at my original dataset. In my notes there were less than 20 patients “meeting the criteria” and that was enough to convince me to set the cohort-trial of the NEJM 1996. You may say that I was lucky but I can reply that all started from careful observations, with the final result achieved after having insisted on endpoints based on that truly original observation.

Pretty awesome, right?

So, my patient has HCC. Now what? How do I get her exception points?

An initial evaluation includes the following:

  1. An evaluation of the number and size of lesions before any therapy that meet T2 criteria (see image below) using a dynamic contrast enhanced CT or MRI.
  2. A CT of the chest to rule out metastatic disease.
  3. A CT or MRI to rule out any other sites of extrahepatic spread or macrovascular involvement.
  4. An indication that the candidate is not eligible for resection.
  5. An indication whether the candidate has undergone local-regional therapy.
  6. The candidate’s alpha-fetoprotein (AFP) level.

After this initial evaluation:

Candidates with HCC lesions are eligible for a standardized MELD exception if they have an alpha-fetoprotein (AFP) level less than or equal to 1000 ng/mL and either of the following:

LEFT: one lesion ≥ 2cm and ≤ 5cm in size
RIGHT: two or three lesions each ≥ 1cm and ≤ 3cm in size

But what if my patient’s lesion or lesions are bigger than that? Can she never get a transplant?

Good news! There are some lesions that can be “down-sized” to T2 lesions and your patient can still be eligible for exception points. These are:

LEFT: one lesion > 5cm and ≤ 8cm in size
MIDDLE: two or three lesions each > 3cm or ≤ 5cm, and a total diameter of all lesions ≤ 8cm
RIGHT: four or five lesions each < 3cm, and a total diameter of all lesions ≤ 8cm

Lesions that were initially larger than these criteria and are then down-sized are not automatically eligible for exception points. These cases must be referred to the board for consideration.

Lastly, what if my patient’s AFP is >1000 ng/mL?

If your patient has an AFP of >1000 ng/mL, but her lesions fall within T2 criteria, you can treat with locoregional therapy. If the AFP drops to <500 ng/mL with treatment and stays there, the patient can be eligible for exception points.

So my patient meets all of the criteria for exception points! Can she be transplanted right away?

No. There is a mandatory 6-month waiting period after patients are deemed eligible for MELD exception points for HCC.

Historical Context:

The MMaT-3 rule was only recently implemented by UNOS. It used to be that exception points were rewarded and then increased every three months that the patient was on the waiting list. This, however, may have over-prioritized those with HCC over those with other etiologies of cirrhosis for transplant. To address this issue, UNOS first required a mandatory 6-month wait period before receiving exception points after studies demonstrated better post-transplant survival in areas with longer waiting lists. This, however, was not enough to address the prioritization of transplant for HCC. The MMaT-3 policy was instituted in order to address this issue. It does seem that the combination of this policy plus the change to acuity circles for transplantation, has led to a decrease in the number of transplants being performed for HCC over the last year. If patients with HCC are not being transplanted as quickly, it is likely that they will need more loco-regional therapy while on the waitlist to keep them within T2 criteria.

But you may ask, why wait 6 months? What is the impact?

In the older system for MELD exception points awarded for HCC, there was concern raised about geographic disparity and optimal prioritization of candidates for transplant. Because patients with HCC were historically awarded an exception score of 22 immediately upon listing without a wait period, many patients with HCC were transplanted very quickly and ahead of other patients with different etiologies of liver disease. This study demonstrated that a delay of 6-9 months would help to eliminate geographic variability in the discrepancy between HCC and non-HCC transplant rates and allow for more equal access to organs. Additionally, waiting 6-months allowed for observation of both particularly aggressive tumor biology that would increase the rate of recurrence post-transplant and for very slow-growing HCCs that may not need to be rushed to transplant.

An important study recently published in Hepatology by Nagai et al. (2020) found that when comparing Organ Procurement and Transplantation Network/United Network for Organ Sharing (OPTN/UNOS) data, compared to before the waiting period was implemented, the HCC policy change was actually associated with decreased waitlist mortality/dropout and increased transplant probability after six months. Additionally, the policy decreased (though did not eliminate) regional disparities.

How and why the MELD score may change again?

Though exception points certainly help to capture some conditions that are not adequately addressed by the MELD score, there are still limitations to its complete generalizability.

First, the components of the score are laboratory values that are prone to inter-laboratory variability. Alterations in the prothrombin time caused by laboratory differences, for example, can lead to differences in MELD scores of 3-12 points! Similarly, bilirubin is subject to variation among laboratories.

Then, there is the fact that creatinine is influenced by muscle mass, protein intake, age, sex, and ethnicity. Additionally, large fluctuations in creatinine in the setting of fluid shits and diuretic treatment, do not necessarily correlate to large changes in mortality risk. However, they DO correlate to large changes in MELD score. Additionally caps to the lower and upper limits of creatinine in the MELD calculation, instituted by UNOS may over or under-represent mortality risk.

Lastly, it is unclear whether MELD is similarly predictive of mortality across multiple different etiologies of liver disease. Hepatitis C used to be the primary reason for transplantation, especially at the time that the MELD score was implemented, but with effective direct-acting antiviral therapy, it is no longer. Now, alcohol related liver disease and non-alcoholic steatohepatitis have become increasingly common as reasons for transplant. In a recent study, MELD was found to be less predictive of mortality in alcoholic liver disease and NASH. In fact, in general, the MELD score’s concordance with 90-day mortality has decreased from 0.80 in 2003 to 0.70 in 2015 despite similar scores at waitlist death around 35 across these two time periods.

Because it is our ultimate goal as the transplant community to continue to improve organ allocation, we must constantly assess whether we can develop better models to predict mortality and to determine transplant priority.

Until we have a better model that is adopted by OPTN/UNOS, we will continue on with MELD-Na. We hopefully now all better understand the score and how it works and can work together to continue to improve it!

Leave a Reply

Your email address will not be published. Required fields are marked *