Clinical and technical evidence

A literature search was carried out for this briefing in accordance with the interim process and methods statement. This briefing includes the most relevant or best available published evidence relating to the clinical effectiveness of the technology. Further information about how the evidence for this briefing was selected is available on request by contacting mibs@nice.org.uk.

Published evidence

There are 4 studies summarised in this briefing, including a total of 5,789 people with type 2 diabetes. The evidence includes 2 validation studies (Peters et al. 2019 and Peters et al. 2020), 1 prospective study (Peters et al. 2021) and 1 early discovery study (Bringans et al. 2017).

In addition, there is a further study assessing the potential of the biomarkers used by PromarkerD to predict rapid decline in renal function in people with type 2 diabetes (Peters et al. 2017). There are also analytical validation studies on PromarkerD (Bringans et al. 2020a) and studies on the stability, reproducibility and precision of the assay (Bringans et al. 2020b).

The clinical evidence and its strengths and limitations are summarised in the overall assessment of the evidence.

Overall assessment of the evidence

The evidence base for PromarkerD is limited and predominantly comes from validation studies. The studies have relatively large sample sizes and one of the validation studies included people from 30 different countries. Despite this, one of the limitations highlighted by study authors and clinical experts was the lack of generalisability of results. This is because of the relative lack of representation of different family backgrounds across the evidence base, particularly those that are at high risk of developing diabetic kidney disease (DKD). One comparative study is included in the evidence base (Peters et al. 2021) comparing the PromarkerD test to the standard care tests used in clinical practice. Further external validation of the technology through comparative studies in a UK context including a wide range of patient groups would be beneficial. Three of the publications cited in this briefing assessed patients from the same study cohort. The time horizon used in the studies is approximately 4 years, which is a relevant prognostic time horizon for people with type 2 diabetes.

Peters et al. (2020)

Study size, design and location

A multicentre validation study predicting renal function decline in 3,568 people with type 2 diabetes. The study was done at 667 centres in 30 countries.

Intervention and comparator

PromarkerD, no comparator.

Key outcomes

Mean baseline estimated glomerular filtration rate (eGFR) was 77 ml/min/1.73 m2. 16.5% of people had renal impairment, classified as having an eGFR lower than 60 ml/min/1.73 m2. 1,351 people (38%) had chronic kidney disease at baseline, defined by a composite of eGFR and urine albumin to creatinine ratio (uACR), eGFR lower than 60 ml/min/1.73 m2 and/or uACR higher than 30 mg/g. Excluding those with renal impairment, 926 people (31.1%) developed chronic kidney disease during the 4‑year follow-up period. During follow up, 564 people (16%) suffered a decline in eGFR of more than 30%.

The prognostic score for PromarkerD was significantly associated with predicting incident chronic kidney disease (p<2.8×10-47). Moderate-risk and high-risk scores were increasingly prognostic for incident chronic kidney disease; odds ratio (OR) 5.29 (4.22 to 6.64) and OR 13.52 (10.69 to 17.11) respectively. The prognostic score was also significantly associated with an eGFR decline of more than 30%; OR 1.13 (1.04 to 1.24). PromarkerD provided 60.6% sensitivity and 82.6% negative predictive value at the 10% cut-off, and 94.0% specificity and 73.9% positive predictive value at the 20% cut-off for predicting 4‑year risk of developing chronic kidney disease. The test performed poorly in differentiating people with rapid eGFR decline from those with lesser declines.

Strengths and limitations

A significant strength of the study was the large sample size. The study included people with type 2 diabetes using a global multicentre approach, independent of the Australian population used to develop the test. This study also describes the development of PromarkerD. One of the limitations highlighted by the authors was that 81% of participants were from a white family background, which limits the generalisability of the PromarkerD test to other family backgrounds. Only baseline clinical and biomarker data was used to predict outcomes; subsequent changes in biomarker concentrations during the follow-up period were not considered.

Peters et al. (2019)

Intervention and comparator

PromarkerD, no comparator.

Key outcomes

The study population was separated into 2 groups: the development cohort (n=345) and the validation cohort (n=447). During a mean follow up of 4.2 years, 39 people (9.8%) in the validation cohort developed DKD and 24 people (5.4%) experienced more than a 30% decline in eGFR.

The predictive performance of PromarkerD was assessed. The model for incident DKD had the highest predictive ability to discriminate between people who did and did not develop DKD during follow up. In the development and validation cohorts, the concordance was 0.89 and 0.88, respectively. PromarkerD provided 86.1% sensitivity at the 10% cut-off, and 84.7% specificity at the 20% cut-off to predict 4‑year risk of developing DKD. For an eGFR decline of 30% or more, the concordance was 0.81 and 0.73 in the development and validation cohorts, respectively.

Strengths and limitations

The study employed a prognostic time horizon of over 4 years that is relevant to people with type 2 diabetes. A limitation of the study was that baseline clinical and biomarker data was used to predict outcomes, but subsequent changes in biomarker concentrations or diabetes management were not considered. The authors stated that additional external validation across different clinical settings and populations is needed to fully realise the generalisability of the predictive models.

Peters et al. (2021)

Intervention and comparators

PromarkerD, compared with eGFR and uACR (standard care).

Key outcomes

The study incorporated the Kidney Disease: Improving Global Outcomes (KDIGO) risk classes for adverse outcomes, which are based on eGFR and albuminuria measurements. The KDIGO categories give the risk of chronic kidney disease progression, morbidity and mortality. At baseline, participants were classified by PromarkerD as low (63%), moderate (13%) or high risk (24%), and by KDIGO as low (58%), moderate (31%), high (7%) or very high risk (4%) for renal decline within 4 years. Of the 497 people in KDIGO low-risk category with normal kidney function, 45 (9%) developed incident DKD within 4 years and would have been missed by standard care tests. PromarkerD classified 38 (84%) of these people as moderate or high risk, flagging them for early intervention and closer monitoring of disease. In addition, 354 out of 361 (98%) people with low-risk PromarkerD results did not develop incident DKD. Of the people who developed the outcome, 84% had moderate or high-risk PromarkerD scores.

During 4.2 years of follow up, 107 people (12.5%) experienced a decline in renal function. Higher PromarkerD risk scores had a stronger association with renal decline (OR 3.26) compared with lower eGFR and higher uACR (OR 2.63 and 1.21, respectively). PromarkerD moderate and high-risk scores were increasingly prognostic for renal decline (OR 8.11 and 21.34, respectively) compared with low-risk scores (p<0.001). PromarkerD has significantly higher predictive performance (concordance of 0.88) compared with standard care tests (eGFR only, concordance of 0.82, uACR only, concordance of 0.63, eGFR + ACR, concordance of 0.82) for predicting decline in renal function within 4 years (p<0.001).

Strengths and limitations

The study used a follow-up period of 4 years, which was useful to compare baseline risk scores with outcomes seen at the end of the follow-up period. As well as comparing PromarkerD directly with eGFR and uACR tests, both individually and combined, test scores were also compared with the KDIGO risk classifications. The study was funded by the manufacturer.

Bringans et al. (2017)

Intervention and comparator

PromarkerD, no comparator.

Key outcomes

The diagnostic performance of the biomarker model was assessed alongside the standard care uACR and eGFR diagnostic tests. uACR data was found to have a true positive rate of 73% and a false positive rate of 40% when diagnosing eGFR of less than 60 ml/min/1.73 m2. In the opposite analysis, eGFR data had a true positive rate of 26% and a false positive rate of 8% when used to diagnose uACR less than 3 mg/mmol. The biomarker eGFR model had an improved true positive rate (88% versus 73%) and a reduced false positive rate (32% versus 40%) over the standard care uACR test for diagnosing eGFR of less than 60 ml/min/1.73 m2. The biomarker uACR model had an improved true positive rate (52% versus 26%) but a poorer false positive rate (15% versus 8%). The diagnostic odds ratios for the eGFR and uACR biomarker models were significantly better than those of the standard care tests (eGFR 14.9 versus 4.0, uACR 6.0 versus 4.0).

Strengths and limitations

The study provides an outline of the methods used to determine the final biomarkers for the PromarkerD test. A significant strength of the study is that the biomarker tests were compared with standard care tests in a large sample of people with type 2 diabetes.

Sustainability

The company does not claim any sustainability benefits of this technology.

Recent and ongoing studies

No ongoing or in-development trials were identified.