2012: The Year of Failed Surrogate Endpoints Reply

Tags: , , , , , , ,

By C. Michael Gibson, M.S., M.D.

2012 will go down as the year of failed surrogate endpoints.  The lack of clinical improvements associated with lowering HDL and increasing platelet inhibition are two notable examples.  What follows comes from my textbook chapter on surrogate endpoints on WikiDoc.org .

The Advantages of a Surrogate Endpoint

The assessment of “hard” primary clinical endpoints (such as death and heart attack) often requires large long-term clinical trials which can be quite expensive. The use of surrogate endpoints can allow trials to evaluate the efficacy of a new drug or device more rapidly, more efficiently and more inexpensively.

The Disadvantages of Surrogate Endpoints

There are several potential disadvantages of a surrogate endpoint.

1. The surrogate endpoint may intuitively be hypothesized to be related to a “hard endpoint” such as death or heart attack, but may not be.

2. While a surrogate endpoint may be related to a “hard endpoint” such as death or heart attack, it is not clear that a reduction in the surrogate endpoint will lead to an improvement in the “hard endpoint” in death or heart attack. Anti-diabetic agents have been shown to reduce long term glucose (Hemoglobin A1c or HbA1C). It was hypothesized that more intense glucose control (a reduction in HbA1C) would be associated with a lower rate of death and heart attack. However, despite lowering of HbA1C, rosiglitazone or Avandia was not associated with a reduction in death and MI, but with an increase in the RECORD trial [5] [6].

In another example, class III antiarrhythmic agents were associated with suppression of premature ventricular contractions or PVCs, but were associated with a higher rate of death.

3. While a surrogate endpoint may be related to a “hard endpoint, it may be an acausal association (the surrogate may not lie in the causal pathway to the “hard endpoint” and changing the surrogate endpoint may not change the “hard endpoint”.)

4. The agent may reduce the surrogate endpoint, but due to off target toxicity, may increase the risk of “hard endpoints” such as death or MI. Lower HDL is associated with a higher risk of adverse cardiac outcomes, Torcetrapib raises HDL and should therefore improve clinical outcomes, however, Torcetrapib administration was found to be associated with a higher rate of adverse clinical outcomes. It was felt that the potential benefit of Torcetrapib was reversed due to off target toxicity of a slight increase in blood pressure associated with Torcetrapib administration.

5. The relationship between the surrogate endpoint and the “hard endpoint” may be non-linear or may be a threshold effect. For example, in antiplatelet agent studies, it is unclear if ever greater levels of inhibition of platelet aggregation are associated with ever greater reductions in adverse outcomes, or if one must achieve just a certain “threshold” level of inhibition to improve outcomes.

There have been a number of instances when studies using surrogate markers have been used to show benefit from a particular treatment, but later, a repeat study looking at endpoints has not shown a benefit, or even a harm.[7]

Examples of Surrogate Endpoints

Examples of surrogate markers in current practice include:

  • Total cholesterol
  • Low density lipoprotein (LDL)
  • Coronary blood flow
  • Fragmented blood cells are a surrogate marker for organ failure or stroke in TTP
  • The S-phase duration, may be used as a surrogate marker for breast cancer occurrence;
  • CD4 count is a surrogate marker for death from HIV infection.
  • Tumor shrinkage
  • CEA levels in colon cancer trials
  • Prostate specific antigen PSA in prostate cancer.

Evaluation of a Surrogate Endpoint

There are multiple criteria that have been proposed to evaluate surrogate endpoints.

Does the Surrogate Lie in the Causal Pathway of Disease, or is it Simply a Prognostic Marker?

Examples of this criteria include CEA in colon cancer and PSA in prostate cancer. As stated by Thomas Flemming “While CEA and PSA are not the mechanism through which the disease process induces increased risk of the clinical-efficacy outcomes, so it is questionable whether treatment-induced changes in these markers could be relied upon to accurately predict treatment-induced effects on the clinical endpoints.[4]”

Are there Multiple Causal Pathways Present?

If there are multiple causal pathways present, what is the relative strength of the proposed surrogate relative to that of other pathways. This may alternatively lead to an over or under estimate of the impact of the surrogate marker.

Is there “Off target” Toxicity Associated with Altering the Surrogate Endpoint?

It is possible that the intervention being test could be associated with other deleterious mechanisms of action that are independent of its intended therapeutic effects[4]. Very often, because such effects are unintended, they are unanticipated, unrecognized, and unrecorded.The classic example of this is the hypertension caused by torcetrapib, or the pro-arrhytmic effects of class III antiarrhythmia agents as demonstrated in the CAST trial[8].

Proposed Criteria for a Surrogate Endpoint

One investigator, Ross Prentice, proposed that the following two conditions must be meant[9]:

  1. The biological marker must be correlated with the clinical endpoint; and
  2. The marker must fully capture the net effect (on target efficacy and off target hazards) of the intervention on the clinical-efficacy endpoint.

From my perspective, a surrogate may be clinically relevant if the following three criteria are satisfied:

  1. The drug or device improves the surrogate
  2. Improvement in the surrogate is related to an improvement in a hard clinical endpoint
  3. The same drug or device improves the hard clinical endpoint

In the current cost constrained era there may be a temptation to utilize surrogate endpoints.   The failures of the past year,  however, provide a cautionary note regarding the hazards of surrogate endpoints.

References

  1. ↑ Cohn JN (2004). “Introduction to Surrogate Markers“. Circulation 109: IV20–1. American Heart Association. PMID 15226247. Retrieved on 2007-01-10.
  2. ↑ Controlled Clinical Trials 22:485–502 (2001))
  3. ↑ Naccarelli GV, Dougherty AH, Wolbrette D, Wiggins S (1991). “A critical appraisal of the cardiac arrhythmia suppression trial (CAST)”. Applied Cardiopulmonary Pathophysiology : ACP 4 (1): 9–16. PMID 10147539. Retrieved on 2010-10-31.
  4. ↑ 4.0 4.1 4.2 Fleming TR (2005). “Surrogate endpoints and FDA’s accelerated approval process“. Health Affairs (Project Hope) 24 (1): 67–78. doi:10.1377/hlthaff.24.1.67. PMID 15647217. Retrieved on 2010-11-21.
  5. ↑ Home PD, Pocock SJ, Beck-Nielsen H, Curtis PS, Gomis R, Hanefeld M, Jones NP, Komajda M, McMurray JJ (June 2009). “Rosiglitazone evaluated for cardiovascular outcomes in oral agent combination therapy for type 2 diabetes (RECORD): a multicentre, randomised, open-label trial“. Lancet 373 (9681): 2125–35. doi:10.1016/S0140-6736(09)60953-3. PMID 19501900. Retrieved on 2010-10-31.
  6. ↑ Nissen SE (March 2010). “Setting the RECORD Straight“. JAMA : the Journal of the American Medical Association 303 (12): 1194–5. doi:10.1001/jama.2010.333. PMID 20332408. Retrieved on 2010-10-31.
  7. ↑ Psaty BM, Weiss NS, Furberg CD, et al. (1999). “Surrogate end points, health outcomes, and the drug approval process for the treatment of risk factors for cardiovascular disease”. JAMA 282: 786–790.
  8. ↑ Echt DS, Liebson PR, Mitchell LB, Peters RW, Obias-Manno D, Barker AH, Arensberg D, Baker A, Friedman L, Greene HL (March 1991). “Mortality and morbidity in patients receiving encainide, flecainide, or placebo. The Cardiac Arrhythmia Suppression Trial“. The New England Journal of Medicine 324 (12): 781–8. doi:10.1056/NEJM199103213241201. PMID 1900101. Retrieved on 2010-11-21.
  9. ↑ Prentice RL (April 1989). “Surrogate endpoints in clinical trials: definition and operational criteria”. Statistics in Medicine 8 (4): 431–40. PMID 2727467. Retrieved on 2010-11-21.

Leave a comment