WO2013122469A2

WO2013122469A2 - Prediction of the efficacy of a medicament on renal or cardiovascular clinical outcomes

Info

Publication number: WO2013122469A2
Application number: PCT/NL2013/050098
Authority: WO
Inventors: Hiddo Jan LAMBERS HEERSPINK; Dick De Zeeuw; Diederick Egbertus Grobbee
Original assignee: Rijksuniversiteit Groningen; Academisch Ziekenhuis Groningen; Umc Utrecht Holding B.V.; Stichting Top Institute Pharma
Priority date: 2012-02-17
Filing date: 2013-02-18
Publication date: 2013-08-22
Also published as: WO2013122469A3

Abstract

For obtaining a prediction of the long-term efficacy of a medicament on reducing a risk of a renal clinical outcome or a cardiovascular clinical outcome, a multivariate risk relation is determined from values (8) of a plurality of biomarkers and the occurrence in the past (11) of said clinical outcome. The biomarkers include systolic blood pressure, ratio between albumin and creatinine contents in urine, and potassium, Hb A1c, and total cholesterol contents in blood. The risk relation is applied to reference biomarker values (16, 18a) for determining a reference estimate of a risk of future occurrence of that clinical outcome (2a) and to medicament affected biomarker values after a short-term treatment period (18b) for determining a medicament affected estimate of a long-term risk of future occurrence of that clinical outcome (2b). The long-term medicament efficacy estimate (19) is determined from the reference risk estimate and the medicament affected risk estimate.

Description

TITLE: Prediction of the efficacy of a medicament on renal or cardiovascular clinical outcomes

FIELD AND BACKGROUND OF THE INVENTION

The invention relates to a method and a computer program product for obtaining an estimate of the efficacy of a medicament in reducing the risk of the occurrence of a clinically meaningful renal or cardiovascular (CV) outcome (e.g. myocardial infarction or dialysis) during a future treatment period.

The current drug development and registration practice in the area of renal and CV diseases targets a medicament towards a risk marker known to be causally related to the risk of CV and/or renal morbidity or mortality, the so called on-target risk marker. For example, an antihypertensive drug or antidiabetic drug is targeted towards reducing blood pressure or towards lowering HbAlc.

However, the ultimate evaluation of the medicament's efficacy is not based on its short-term on-target effect (e.g. on blood pressure or HbAlc), but based on the medicament's efficacy in reducing the risk for long-term renal or CV morbidity or mortality. The change in the on-target risk marker value brought about by the medicament (as measured during short-term studies generally of a duration of up to six months) is used as a measure of the medicament's efficacy and as an indication that a reduction of the risk of an adverse cardiovascular and/or renal outcome over a future time period is likely to occur. In this context, a renal outcome is typically defined as one or more of the following events: halving of the glomerular filtration rate, chronic dialysis (> 2 weeks), renal transplantation, or renal death and CV outcome is typically defined as one or more of the following events: myocardial infarction, stroke, hospitalization for heart failure, or CV death.

If the new medicament is promising in the sense of effectively causing a change in the on-target risk marker and the side-effect profile is acceptable, the medicament is registered. To verify that the medicament is indeed reducing the risk of clinically meaningful outcomes, the short-term study is followed by one or more long-term studies, performed before or after

registration. Such long-term studies (generally up to three / four years of duration) require large patient populations, and large financial and human resources. Over the years numerous medicaments with promising substantial effects on the on-target risk marker have failed to afford long-term renal or CV protection. ^1-5 (Reference numerals in superscript refer to the list at the end of the detailed description.) This has led to drug attrition in late stages of drug development and even marketing withdrawal of already registered drugs.

Such misestimates of the effects of a new medicament on clinically meaningful renal or CV events have major consequences for several stakeholders in drug development and registration:

^■ Patients are exposed to ineffective or even harmful medicaments.

■ For the pharmaceutical industry a high attrition rate involves high costs with no return on investments.

■ Registration authorities perform regulatory review assessments for a

medicament that is ultimately withdrawn.

^■ Trust in drug development as well as willingness of patients to voluntarily take part in clinical drug trials suffers.

In the context of the high incidence and prevalence of renal and CV morbidity and mortality and increasing complexity of developing new

medicaments, there is a strong need for quantitative accurate estimation of the effect of a new medicament on changing the risk of renal or CV morbidity, preferably during the earliest stages of development. Such early estimation may guide the developing process of a new medicament and decisions as to whether further large expensive trials should be pursued. It would therefore be advantageous to possess a tool that allows accurate estimation of the efficacy of a new medicament on renal or CV outcomes. Having a flexible tool that can also be used to identify sub-populations of patients that may additionally benefit from the new medicament would be of particular importance in the era of personalized medicine and tailoring the right drug to the right patient. SUMMARY OF THE INVENTION

It is an object of the present invention to provide an accurate

quantitative prediction of the long-term renal and cardio vascular protection efficacy of a medicament. According to the invention, this object is achieved by providing a method according to claim 1.

By applying a multivariate risk relation between multiple biomarkers and a predetermined set of clinical renal/CV outcomes (as obtained from previous observations) to biomarkers values measured in a different population of subjects with and without treatment with a new medicament, provides a very accurate estimation of the efficacy of that medicament on the predetermined set of clinical renal/CV outcomes for that population.

Availability of a large background population is a prerequisite to establish an accurate biomarker - renal/CV risk relation and allows more flexibility with respect to the use of sub-populations stratified or selected in accordance with characteristics of the subject or population of subjects for which a prediction estimate is desired without narrowing down the population on which the risk relation is based to a too small population.

The medicament of which the efficacy is estimated may be a specific formulation of one or more substances in a particular form and/or dosage to be administered or, more in general, a drug in the sense of a substance for the treatment or prevention of a renal and/or CV clinical outcome as a component of a medicament.

The invention can also be embodied in a computer program product according to claim 14, which is specifically arranged for carrying out the method according to the invention. Particular elaborations and embodiments of the invention are set forth in the dependent claims.

Further features, effects and details of the invention appear from the detailed description and the drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 is a schematic flow chart representation of an example of a method according to the invention;

Figure 2 is a graphic representation of changes in biomarker values during six months treatment in two trials: treatment with the Angiotensin Receptor Blocker losartan in the RENAAL trial and treatment with irbesartan in the IDNT trial.

Figures 3A and 3B are graphic representations of observed and predicted long-term relative renal (Fig. 3 A) and CV (Fig. 3B) risk changes (%) based on individual and multiple measured biomarker values. The actual observed treatment effect is indicated by a solid line. The predicted treatment effect based on individual and multiple bio marker values are shown by the vertical bars. The predictions are based on risk relations calculated from the RENAAL trial and applied to the baseline and month 6 measured biomarker values of the losartan and placebo arm of the RENAAL trial.

Figures 4A and 4B are further graphic representations of observed and predicted long-term relative renal (Fig. 4A) and CV (Fig. 4B) risk changes (%). The risk relation is developed in the RENAAL trial and validated in the IDNT trial by applying the RENAAL developed risk relation to the baseline and month 6 measurements of the irbesartan and placebo arm of the IDNT trial.

Figures 5A-5D are further graphic representations of observed and predicted long-term relative renal (Figs. 5A and 5C) and CV (Figs. 5B and 5D) risk changes (%) based on individual and multiple measured biomarker values. The actual observed treatment effect is indicated by the solid line. The predictions are based on risk relations calculated from the IDNT trial and applied to the baseline and month 6 measurements of the placebo and irbesartan treatment arm of the IDNT trial (Figs. 5A and 5B). The IDNT developed risk relation is validated in the RENAAL trial by applying the risk relation to the placebo and losartan treatment of the RENAAL trial (Figs. 5C and 5D).

DETAILED DESCRIPTION The invention is first described with reference to Fig. 1 which illustrates how, in an example of a method according to the present invention, use is made of measurements and observations, data storage and data processing.

It is desired to obtain a prediction of the efficacy of administering a new medicament over a future treatment period of time, for instance in a long-term trial 1 on reducing a risk of a future occurrence of a clinical meaningful predetermined renal or CV outcome. More specifically, the difference is estimated between the risk 2b of occurrence of a predetermined clinical renal/CV outcome in the treatment arm 2b and control arm 2a of a new long- term trial 1. The control arm can be placebo treatment or an active comparator medicament. In Fig. 1 the control arm is represented by a placebo treatment arm.

At the start of the short term trial 3, baseline values 4 of multiple biomarkers are measured. The biomarkers measured in the present example include systolic blood pressure ratio between albumin/creatinine contents in urine and total cholesterol contents in blood. These biomarkers constitute biological or physical variables reflecting a subject's cardiovascular health. The measured biomarkers may also include other biomarkers, such as uric acid, Body Mass Index and/or potassium, HbAlc (in particular for diabetes patients), haemoglobin, calcium, phosphate and/or albumin contents in blood. A particularly accurate and robust prediction can be obtained if the biomarkers on which the risk relation is based and which are measured and used for establishing the prediction also include potassium, haemoglobin, calcium and albumin contents in blood. Further improvement of the accuracy and robustness can be achieved by also including the biomarkers uric acid and/or phosphate contents in blood. During a final stage (preferably at the end) of the short term trial 3, end of treatment values 5 of the plurality of biomarkers are measured. The baseline and end-of-treatment biomarker values 4, 5 are stored in a database 6 of the new trial.

Biomarker values 8 and clinical outcomes 11, including points in time of occurrence of the respective outcomes, measured and observed in previous long-term intervention trials 7 have been measured and stored for each subject in a previous trials database 9. Such a long-term observational follow-up generally lasts 3-5 years. The previous trials database 9 constitutes an aggregated trial database containing data gathered from measurements during various intervention trials 7.

From database 9, stored data 12 representing the biomarker values 8 of the subjects of the population of previous trials 7 and data 13 representing the clinical outcomes 11 for the subjects of that population are utilized. At step 14, a multivariate risk relation is determined from the previous values of the plurality of biomarkers 8 and the occurrences of the clinical outcome 11, for instance using a multivariate Cox proportional hazard model.

At step 15, a baseline estimate of a risk of occurrence of the clinical outcome for the population enrolled in the new short-term trial is calculated from the multivariate risk relation. At step 17a and 17b, the average risk of a renal or CV outcome during the period of new long term trial 1 is calculated for the population at the end of the short-term trial for the control arm and active treatment arm separately. The average risk is based on the biomarker values obtained at the end of the short term study 5 and are calculated based on the risk relation between biomarker values 8 and observed clinical outcomes 11 stored in database 9. Subsequently, the risk changes are calculated at step 19. The differences between end of treatment and baseline risk are calculated for both the active treatment arm and the control arm. The difference between the risk change in the treatment arm and the control arm constitutes the estimate of new medicament's efficacy on the long-term predetermined clinical renal and CV outcomes.

By calculating the predicted difference between the risks 2a, 2b of a clinical outcome in a future long-term trial 1 by applying a multivariate relation between biomarkers and clinical renal/CV outcomes (obtained from previous observation and intervention trials) to baseline and end of study biomarker values of a population being treated for a short period of time with a new medicament, a very accurate prediction of the change of the risk of clinical renal/CV outcome in a future long-term trial 1 can be obtained. The methodology can for instance be applied to various populations of patients with type 2 diabetes (e.g. subjects with good, moderate or poor renal function) since an extremely large amount of background data obtained from large and heterogeneous populations of patients with type 2 diabetes is available. The simplicity of the biomarker value - risk function, allows easy stratification and selection of a population from the background data that has similar

characteristics of the population enrolled in a new short-term trial. In doing so, an accurate long-term renal/CV drug efficacy estimate can be obtained for a population with similar characteristics as the short-term trial.

It was found that in this field of medical treatment, the medicaments have little effect on the risk relation between multiple on-target and off-target biomarkers and the risk of occurrence of clinical outcomes. Accordingly, the relation between biomarker values and clinical outcomes from previous trials can be applied to baseline and medicament affected biomarker values of the new short term trials.

The prediction of the long-term risk reduction achieved with the medicament under investigation, preferably applies to a treatment period 1 that ends at least one year, and more preferably at least two or at least four years, after the short term trial period 3, since observations on the occurrence of clinical occurrences are typically available from and desired for trials of such durations.

The method according to the invention may also be used for predicting whether a medicament is effective for a particular patient, for instance by administering the medicament to that patient and comparing the efficacy prediction for that patient with the efficacy prediction for another individual or another drug within the same class of drugs. Such predictions may offer the physician and the patient a better indication on the prescribed drug effect on long-term outcomes. This makes the method according to the invention particularly relevant for the patient-clinician dialogue and will aid to guide the intensity of renal or CV protective drug therapy.

The method can also be repeated for a particular patient or population of patients with different dosages of the same medicament, wherein the efficacy is predicted for each dosage of a medicament and long-term trials or clinical administering of the medicament is subsequently carried out with the dosage for which the largest risk reduction was predicted. The dosages may for instance be defined in terms of quantities of medicament per unit of time or in terms of a titration to a biomarker level or to an aggregate of multiple biomarker levels.

A category of patients for which the prediction of the long-term risk has been found to be particularly accurate are patients with type 2 diabetes and signs of renal impairment.

EXPERIMENTAL EXAMPLE

Study design

For developing the multivariate risk relationships, data were used from the Reduction of Endpoints in NIDDM a trial of losartan, an Angiotensin II Antagonist (hereinafter the RENAAL trial) and a trial of irbesartan

(hereinafter the IDNT trials). The rationale, study design and outcomes for these trials have been previously published and were similar ⁶"⁹. In brief, the overall aim of the trials was to assess the impact of an Angiotensin Receptor Blocker (ARB) on hard renal (primary endpoint) and CV outcomes (secondary endpoint) by testing losartan 100 mg/day vs. placebo in the RENAAL trial and irbesartan 300 mg/day vs. placebo in the IDNT trial. The IDNT trial also included a Calcium Channel Blocker arm which was not used in the present analysis. Inclusion criteria for both trials were presence of type 2 diabetes, nephropathy, and age between 30 and 70 years. Subjects with insulin dependent diabetes or renal disease not related to diabetes were excluded in both trials. All subjects gave written informed consent. Both trials were approved by local medical ethics committees and conducted according to the guidelines of the declaration of Helsinki.

Measurements

In both RENAAL and IDNT, renal and CV biomarkers, hereinafter referred to as risk markers, were measured at baseline and at 6 months intervals thereafter. All risk markers collected at baseline and month 6 were used to create a "PRE score" constituting an estimate of the change of the risk based on the change in the value of the respective risk marker or combination of risk markers. We selected all measured risk markers at month 6 because we did not know a priori which risk markers would change during ARB therapy and secondly to exclude any potential bias as a result of risk marker selection. Changes in both on-target and off-target risk markers after ARB treatment were calculated as the difference between the baseline and the 6-month value. 6-month values were chosen because most parameters were available at 6- month and ARB treatment effects were considered fully present. Since total cholesterol, haemoglobin, serum albumin, calcium, and phosphate were not measured at month 6 in the RENAAL trial, 12-month values were used for these risk markers in the RENAAL trial.

Renal and CV outcomes

The primary outcome for the present analysis was defined as a composite of a confirmed doubling of serum creatinine from baseline (DSCR) or end-stage renal disease (ESRD). The latter was defined as chronic dialysis or renal transplantation. The secondary CV outcome was another endpoint for the present study which was defined in both trials as the composite of myocardial infarction, stroke, hospitalization for heart failure,

revascularization procedures or death related to CV disease. All renal and CV events were adjudicated by an independent blinded committee using rigorous definitions. Model development

Parameter risk response scores were developed by estimating the risk relation between single or multiple risk markers and renal or CV outcomes in the placebo group of the RENAAL trial. The single and multiple parameter risk response outcome scores were subsequently applied to the baseline and 6- month measurements of the ARB treatment arm to predict renal or CV risk at both time points. The difference in the estimated risk at these time-points, adjusted for the difference in estimated risk in the placebo arm, indicates the long-term renal or CV risk change conferred by ARB treatment. To test the validity of this approach, the single and multiple parameter risk response outcome scores were compared with the actual observed renal or CV outcomes of the trials. Any model tends to show too optimistic performance when applied to the dataset from which it is developed. The risk response scores were externally validated by developing the scores in the RENAAL trial and testing them in the IDNT trial. Model evaluation

The methodology to develop the PRE score assumes that the association between risk markers at baseline and renal or CV events in the placebo group is similar to the association between single or multiple risk markers at 6- month and renal or CV events during ARB therapy. To verify the validity of this assumption, we determined whether 6 months ARB treatment modified the association between risk markers and renal or CV events. We detected no interaction between the association between risk markers and renal or CV events and ARB treatment for renal or CV outcomes. This indicates that ARB treatment did not modify the association between single or multiple risk markers and renal or CV events. Imputation of missing data yielded essentially similar results as the main analyses.

Statistical analysis

Mean and standard deviation were provided for 6-month changes in risk markers and statistical significance for the between group difference was determined based on a two sided t-test. Univariate and multivariate Cox proportional hazard regression was used to determine the relationship between baseline risk markers in the placebo treatment arm and renal or CV outcome. For subjects who experienced more than one renal or CV event during follow-up, survival time to the first relevant endpoint was used in each analysis. Participants were censored either at their date of death or, for those still alive, at the end of follow-up, the date of their last clinic visit before the termination of the trials. The multivariate Cox analysis included the following risk markers, systolic blood pressure, urinary albuminxreatinine ratio

(UACR), potassium, haemoglobin, uric acid, HbAlc, total cholesterol, Body Mass Index, calcium, phosphate, and albumin. Bootstrap methods were applied, repeating the entire modelling with 1000 independent bootstrap samples, to include the variability of the regression coefficients of single and multiple risk scores and renal or CV outcomes. To determine the actual observed effect of losartan or irbesartan on renal/CV outcomes, the

dichotomous treatment variable was used in a Cox regression model and the relative risk reduction was calculated as (1- hazard ratio) multiplied by 100%. Bootstrap methods were used, repeating the entire modelling with 1000 independent bootstrap samples, to calculate the standard error and 95% confidence interval of the difference between the actual observed and predicted treatment effect. The difference between the predicted and observed treatment effect was tested by means of two sided t-tests. Normal distribution of the predicted treatment effect was verified and was log-transformed if required. Time-dependent Cox regression analysis was used to assess the interaction between risk markers at baseline in the placebo group and 6-month risk markers in the ARB treatment group with renal/CV outcomes. Some patients had missing values for relevant baseline or 6-month variables. To assess the impact of missing data we conducted an additional analysis. Missing data were imputed by using linear regression models. Each dependent variable at baseline or 6-month was run in the model with a set of variables with complete measurements (independent variables). These variables served as the predictors for patients who had missing values. A two-sided p value of 0.05 indicated the nominal level of statistical significance. Analyses were conducted with R version 2.10.1 (R Project for Statistical Computing www.r-project.org).

Results

Baseline characteristics between the ARB and placebo groups in the RENAAL trial was were well balanced⁸. Losartan significantly changed multiple off-target renal or CV risk markers beyond blood pressure. Relative to placebo, losartan decreased UACR, total cholesterol, haemoglobin, and uric acid, it increased potassium, calcium, albumin, and body mass index, while it had no effect on HbalC and phosphate (figure 2). Estimated renal and CV treatment effect by single and multiple parameter risk response outcome scores

During 3.4 years of follow-up 489 renal and 515 CV events were recorded in the RENAAL trial. Treatment with losartan resulted in a relative renal risk reduction of 21.8% (95% CI, 6.5 to 34.5%; p=0.007) and 9.2% (95% CI -7.9 to 23.6%; p=0.27) relative CV risk reduction, as represented by the horizontal line in figure 3^A and 3^B, respectively. Single risk markers failed to accurately predict the renal or CV outcome since they either underestimated or overestimated the actual observed drug effect (figure 3^A and 3^B). The multiple PRE score including all risk markers estimated a long-term relative renal risk reduction of 30.1% (95% CI; 10.8 to 49.5%), which came close to the actual observed relative renal risk reduction (p=0.44), and predicted 9.4% (95% CI 1.9 to 17.0%) relative CV risk reduction which was nearly equal to the observed risk reduction (p=0.98 vs. observed relative risk reduction; figure 3^A and 3^B).

External validation of the multiple parameter risk response outcome score

To test the validity of the PRE score we applied it to an external separate trial database, the IDNT trial, to estimate the treatment effect of the ARB irbesartan on renal and CV outcomes. Irbesartan caused similar directional changes in renal or CV risk markers as losartan although the magnitude of these changes varied compared with losartan (figure 2).

When we entered the irbesartan induced changes in multiple renal or CV risk markers in the PRE score, developed in RENAAL, the PRE score estimated a 26.6% (95% Confidence Interval 14.3 to 38.9%) relative renal risk reduction which was nearly similar to the actual observed relative renal risk reduction of 26.0% (6.4 to 41.5%; p=0.95 vs. predicted drug effect; figure 4^A). The PRE score estimated irbesartan's CV treatment effect to be 7.9% (1.3 to 14.5%) which did not differ from the actual observed CV treatment effect of 11.9% (-8.4 to 28.5 %; p=0.67; figure 4^B). Development of the PRE score in the IDNT trial and application to the irbesartan arm of IDNT or losartan arm of the RENAAL trial yielded essentially similar results in that the estimation of the observed treatment effect based on the multiple risk response score outperformed scores based on single risk markers (figure 5^A - 5^D).

REFERENCES

1. James WP, Caterson ID, Coutinho W, et al. Effect of sibutramine on cardiovascular outcomes in overweight and obese subjects. N Engl J ed;363(10):905-17.

2. Blind E, Dunder K, de Graeff PA, Abadie E. Rosiglitazone: a European regulatory perspective. Diabetologia;54(2):213-8.

3. Yusuf S, Teo KK, Pogue J, et al. Telmisartan, ramipril, or both in

patients at high risk for vascular events. N Engl J Med

2008;358(15):1547-59.

4. Mann JF, Green D, Jamerson K, et al. Avosentan for overt diabetic

nephropathy. J Am Soc Nephrol;21(3):527-35.

5. Pfeffer MA, Burdmann EA, Chen CY, et al. A trial of darbepoetin alfa in type 2 diabetes and chronic kidney disease. N Engl J Med

2009;361(21):2019-32.

6. Brenner BM, Cooper ME, de Zeeuw D, et al. The losartan renal

protection study-rationale, study design and baseline characteristics of RENAAL (Reduction of Endpoints in NIDDM with the Angiotensin II Antagonist Losartan). J Renin Angiotensin Aldosterone Syst

2000;l(4):328-35.

7. Rodby RA, Rohde RD, Clarke WR, et al. The Irbesartan type II diabetic nephropathy trial: study design and baseline patient characteristics. For the Collaborative Study Group. Nephrol Dial Transplant

2000;15(4):487-97.

8. Brenner BM, Cooper ME, de Zeeuw D, et al. Effects of losartan on renal and cardiovascular outcomes in patients with type 2 diabetes and nephropathy. N Engl J Med 2001;345(12):861-9.

9. Lewis EJ, Hunsicker LG, Clarke WR, et al. Renoprotective effect of the angiotensin-receptor antagonist irbesartan in patients with

nephropathy due to type 2 diabetes. N Engl J Med 2001;345(12):851-60.

Claims

1. A method for obtaining an estimate of the efficacy of a medicament on reducing a risk of a renal clinical outcome or a cardiovascular clinical outcome over a future treatment period of time, comprising:

obtaining a multivariate risk relation between values (8) of a plurality of biomarkers and the occurrence (11) of said clinical outcome, said

multivariate risk relation having been determined from:

for the subjects of a first population, values (12) of a plurality of biomarkers including systolic blood pressure ratio between

albumin/creatinine contents in urine and total cholesterol contents in blood, measured from said subjects prior to a completed first period of time, and

data (13) on the occurrence of said clinical outcome for said respective subjects during said first period of time;

administering the medicament to a subject or to at least a portion of a second population over a second period of time;

obtaining reference values (16, 18a) of said plurality of biomarkers measured from said subject or second population during an initial stage of the second period of time or from a portion of said population to which said medicament has not been administered;

obtaining medicament affected values (18b) of said plurality of biomarkers measured from said subject or second population during a final stage of the second period of time;

determining a reference estimate of a risk of future occurrence of said clinical outcome (2a) from said reference biomarker values (16, 18a) and said multivariate risk relation; determining a medicament affected estimate of a risk of future occurrence of said clinical outcome (2b) from said medicament affected biomarker values (18b) and said multivariate risk relation;

determining the efficacy estimate (19) from said reference risk estimate and said medicament affected risk estimate.

2. A method according to claim 1, wherein the plurality of biomarkers from which the risk relation is determined and for which reference values (16, 18a) and medicament affected values (18b) are obtained including at least: uric acid, Body Mass Index and/or potassium, HbAlc, haemoglobin, calcium, phosphate and/or albumin contents in blood.

3. A method according to claim 1 or 2, wherein the reference values of said plurality of biomarkers comprise baseline values (16) measured from said subject or second population during an initial stage of the second period of time.

4. A method according to any of the preceding claims, wherein the reference values of said plurality of biomarkers comprise control values (18a) measured during the final stage of the second period of time from a portion of said population to which said medicament has not been administered.

5. A method according to any of the preceding claims, wherein said first population includes subjects to which a placebo has been administered.

6. A method according to any of the preceding claims, wherein said first population includes subjects to which a different medicament than the medicament of which the efficacy is to be estimated has been administered.

7. A method according to any of the preceding claims, wherein the future period of time for which the risk estimates are determined ends at least one year after the second period of time.

8. A method according to any of the preceding claims, wherein the medicament is administered to a type 2 diabetes or nephropathy patient or to a population of type 2 diabetes or nephropathy patients.

9. A method according to claim 8, wherein the first population is a population of type 2 diabetes or nephropathy patients.

10. A method according to any of the preceding claims, wherein the first population is a virtual population at least selected or stratified from a base population of which said biomarker values and clinical occurrence data are available.

11. A method according to any of the preceding claims, wherein the efficacy is estimated for a dosage of a medicament.

12. A method according to claim 11, wherein the dosage of the

medicament is defined in the form of a titration to a level of at least one biomarker.

13. A method according to claim 11 or 12, repeated for different dosages.

14. Computer program product comprising at least one computer readable medium having stored thereon, in non-transitory form:

a computer executable algorithm (14, 15, 17a, 17b, 19) comprising: a multivariate risk relation between values (8) of a plurality of biomarkers and the occurrence (11) of said clinical outcome, said multivariate risk relation having been determined from:

reference input variables for inputting reference values (16, 18a) of a plurality of said biomarkers measured during an initial stage of a second period of time;

medicament affected input variables for inputting medicament affected values (18b) of said plurality of biomarkers measured during a final stage of the second period of time;

a reference risk calculation routine (15, 17a) for calculating a reference risk estimate of a future occurrence of said clinical outcome from said reference values (16, 18a) and said multivariate risk relation;

a medicament affected risk calculation routine (17b) for calculating a medicament affected risk estimate of a future occurrence of said clinical outcome from said medicament affected values and said multivariate risk relation; and

an efficacy calculation routine (19) for calculating the efficacy estimate from said reference risk estimate and said medicament affected risk estimate.

15. A computer program product according to claim 14, further comprising a database (9) containing said previous values (12) of said plurality of biomarkers measured from a base population prior to said completed first period of time and said data (13) on the occurrence of said clinical outcome in said base population during said first period of time and a risk function calculation routine for calculating a risk function for a subject or a populat from a virtual population selected or stratified from said base population.