Gold Nanoparticle Sensors for Detecting Chronic Kidney Disease and Disease Progression

Ophir Marom; Farid Nakhoul; Ulrike Tisch; Ala Shiban; Zaid Abassi; Hossam Haick

Nanomedicine. 2012;7(5):639-650.

Discussion

Identification of CKD & Disease Progression Using the GNP Sensors

We have studied the feasibility of utilizing combinations of GNP sensors and separate sensors to distinguish between:

Early-stage CKD and healthy states
Stage 4 and 5 CKD states
Early-stage (stages 2 and 3) and late-stage (stages 4 and 5) CKD

Excellent distinction was achieved for the first two cases by means of SVM analysis of two separate sets of sensing features. A single sensing feature was sufficient for the distinction between early and late stages in the third case. In principle, it would be possible to use only one sensing feature in all three cases, but scattered experimental data reduce the detection ability and reliability of the results. Only in the third case was sensor S1 so dominant in its capacity to discriminate the given classes that other sensors were not required. Furthermore, the optimal set of sensing features for the identification of early-stage CKD differed from the best feature set for the distinction of advanced and end-stage CKD. This can be understood if one considers the observed changes in the chemical composition of the breath during CKD disease progression. However, the sensing mechanism of chemiresistive GNP layers is still subject of scientific controversy.^[26,40] Therefore, no strict connection can be proposed at the present stage between the discriminative ability of the selected sensors/sensing features and the exposure to certain VOC mixtures.

Sensors S1–S4 were based on chemiresistive layers of different types of GNPs, which differ in their ability to absorb certain classes of VOCs. However, the organic ligands of the GNPs provide only a moderate chemical selectivity, so that the sensors are cross-reactive. The ligands are selected based on their ability to absorb certain (classes of) small VOCs that are typically found in exhaled breath as metabolic products. The changes in chemical composition between healthy and early-stage CKD differ from the changes in chemical composition between advanced and end-stage CKD. Hence, distinct combinations of sensing features for the different chemistries would provide the optimal results in these two cases. Note that sensors S1–S3 have been used previously for the identification of the breath VOC patterns of different types of cancer.^[13,30,39] This is because metabolic breath VOCs may appear at different concentration ratios in the breath of healthy persons, CKD patients, and patients suffering from other diseases such as cancer, as discussed later. Therefore, the VOC patterns of CKD and cancer can indeed be identified with different combinations of sensing features that are read out from the same or similar sensors.

The SVM classification revealed 77% sensitivity, 80% specificity and 79% accuracy for the early-stage detection of CKD states. The obtained values fulfill the criteria for a good diagnostic method. The GNP breath test could be applied by general practitioners in nonspecialist settings to ensure the speedy referral of a new patient to a specialist, who would initiate an appropriate therapeutic approach. Detecting progression from stage-4 to stage-5 is equally important, because it usually marks the onset of dialysis treatment. Cross-validation (Figure 1B) showed that the progression to ESRD could be identified with high sensitivity (75%), specificity (92%) and accuracy (85%). The sensitivity, specificity and accuracy of detecting the progression from early to late-stage disease using a single sensing feature were 75, 77 and 76%, respectively. Note that these are the preliminary results of a pilot study with a limited sample size that was aimed at providing a proof-of-concept. Large-scale clinical trials are time consuming and are, therefore, often preceded by a small pilot study to test the feasibility of the studied method. An extended clinical study with a larger study population is currently underway to verify the obtained sensitivity and specificity.

Age, gender, diet, smoking habits, medication and exposure to hospital atmosphere are known to affect the chemical composition of the exhaled breath.^[23,24,26] Ideally, the study population should be matched in terms of these confounding factors. We have previously shown that the GNPs used by us have very little sensitivity to VOCs stemming from the above confounding factors.^[13,39] Therefore, it is permissible to relax (at least, in part) the criteria for the compared groups. Table 1 shows that the population in the three studied cases was nevertheless relatively well matched in terms of most of the confounding factors. The compared groups are age matched in the first two cases and gender-matched in third case. Most of the volunteers were nonsmokers and the populations were well matched with respect of their smoking habits in all three cases. All CKD patients in this study were treated with calcium channel blockers, diuretics and/or α-blockers. The healthy controls did not receive these medications. Hence, the study population was matched in terms of medication in the last two cases. Finally, all volunteers, including the healthy controls, were recruited in the same hospital ward and were, therefore, exposed to the same confounding hospital atmosphere prior to the test.

Trends in the Chemical Composition of the Breath of CKD Patients

The observed trends with disease progression that are shown in Table 2 and Figure 3 could be explained by CKD-specific metabolic processes and/or with the accumulation of toxins due to loss of kidney function. CKD patients suffer from various types of carbohydrate metabolism disorders, which could affect the breath concentrations of the VOCs in the first two groups. The majority of CKD patients suffer from glucose intolerance, which is mainly attributed to insulin resistance.^[41] Whether insulin resistance is the cause or the consequence of CKD is a subject of debate in the literature, and the lack of data from systematic studies has left the question open.^[41,42] Insulin resistance has been associated with a failure in activating pyruvate dehydrogenase under induced CKD conditions.^[43,44] This means that pyruvate cannot be converted into acetyl-CoA – a precursor for cholesterol that produces isoprene^[24,45] and ketone bodies (such as acetone)^[27,45] as byproducts. Hence, a reduction in the level of acetyl-CoA could explain the gradual decrease of isoprene during CKD advance and the rapid decrease of acetone (a recognized biomarker for diabetes^[24]) in early stages (stages 2 and 3) of the CKD. The increase of acetone in the late stages (stages 4 and 5) could be related to the higher percentage of diabetic patients in those classes that are characterized by higher concentrations of breath acetone.

The irregularity in the pyruvate conversion to acetyl-CoA is consistent with another observation. We have shown that acetoin increases as CKD progresses (Figures 3 & 4). This could be attributed to a process that occurs in CKD patients, which favors the conversion of pyruvate to acetoin and 2,3-butylen glycol over the conversion to acetyl-CoA.^[46,47] However, it should be noted that the amount of acetoin found was just above the detection limit of our GC-MS system, suggesting that the readings in the early stages (healthy to stage 3) might be less accurate. It is possible that the real change of acetoin over the progress of CKD is more gradual. Alternatively, acetoin could be of exogenous origin, because it is a common food additive (sweetener), and its accumulation in the breath of late-stage patients may be due to kidney-function loss.^[48]

The metabolic route of ethylene glycol is linked directly to glycolaldehyde, which is a short-chain aldehyde.^[49] Short-chain aldehydes react with amino groups, which eventually leads to advanced glycation end products.^[50] Advanced glycation end products are associated with age-related disease and could also affect kidneys function.^[50] The accumulation of ethylene glycol could be an indication of an increase in advanced glycation end products, or could be related to the high percentage of diabetic patients as was the case for acetone. Still, ethylene glycol could be of exogenous origin, because it is a known environmental toxin,^[51] and its accumulation during late-stage CKD could be attributed to the failure of kidney function.

Group 3 comprises compounds that show significant increase in concentration in early-stage of CKD (stage 2). We assume that this is related to the continual death of kidneys nephrons, as up to 60% of the nephrons die before an accumulation of creatinine and urea could be observed. Collectively, the compounds in group 3 could be biomarkers for early-stage disease (e.g., for disease-related oxidative stress). For example, the sudden increase of styrene in stage 2 of the disease could be explained by the secretion of styrene by tubular maximum mechanism, which reaches its full capacity at stage 2. On the other hand, the compounds in group 3 could also be of exogenous origin. Styrene is an environmental toxin that causes DNA damage, and hydrocarbons are often released as environmental pollutants by the petrochemical industry.^[52]

The compounds in group 4 show no significant trends that could be related to the disease progression. However, it should be noted that slightly subsignificant changes in the concentration (p-values between 0.06 and 0.08) were observed, which could prove to be meaningful in a future study with an increased sample size.

Note that changes in the concentration of some of the identified compounds were also detected in the breath of cancer patients.^[13,30,39] The detected small molecules are VOCs that are either metabolic products or toxins that are absorbed from the environment and are accumulated in the body due to reduced kidney function. As such, each separate compound may appear not only in the breath of healthy persons and CKD patients, but also in patients having other diseases (e.g., cancer^[13,30,39]). This way, breath samples of patients suffering from distinctly different diseases may have some common constituent compounds, but at different concentration ratios.

1 2 3 4 5

Next Section

Table 1. Clinical and biochemical variables for all tested patients and healthy controls.
Variable	Healthy (n = 20)	Stage 2 (n = 6)	Stage 3 (n = 16)	Stage 4 (n = 12)	Stage 5 (n = 8)
Estimated GFR (ml/min per 1.73 m²)^†	>120	60–90	30–59	15–29	≤15
Gender (% males)	55	100	75	66	100
Average age ± standard deviation	46 ± 16	41 ± 19	54 ± 12	63 ± 9	66 ± 15
Smoking history (current/ex/never)	1/0/19	2/1/3	2/1/13	2/1/9	1/1/6
Diabetes mellitus	2 (10%)	0	6 (37%)	8 (66%)	5 (62%)
Arterial hypertension	7 (35%)	5 (83%)	14 (87%)	11 (91%)	7 (87%)
Bodyweight (kg)	75 ± 15 (range: 45–110)	71 ± 8 (range: 63–86)	82 ± 16 (range: 60–111)	82 ± 18 (range: 55–120)	83 ± 21 (range: 53–120)
Serum creatinine concentration (mg/dl)	0.7–1.1^‡	1.19–1.5	1.32–2.65	2–3.8	4.43–7
Potassium (mEq/l)	3.5–5.1^‡	4.7 ± 0.8	4.7 ± 0.6	4.8 ± 0.5	5.1 ± 0.6
Calcium (mg/dl)	8.5–10.5^‡	9.9 ± 0.1	9.3 ± 0.4	9.8 ± 0.3	9.0 ± 0.1
Phosphorous (mg/dl)	2.5–5.0^‡	3.2 ± 0.4	3.9 ± 0.5	4.2 ± 0.6	4.9 ± 0.6

The biochemical variables were obtained from blood tests performed during the month prior to the breath sampling. All subjects refrained from smoking and drinking coffee for 1h prior to the breath sampling.
†Estimated according to Modification of Diet in Renal Disease formula. Note that the test population did not include African–Americans for whom a slight alteration of the Modification of Diet in Renal Disease formula is required due to differences in the metabolism.
‡Nominal values.GFR: Glomerular filtration rate.

Table 2. Changes in chemical composition of the breath during chronic kidney progression.
Compound name	Significance of the distinction between chronic kidney disease subpopulations (p-value)				Possible origin
Compound name	Healthy vs stage 2	Stage 2 vs 3	Stage 3 vs 4	Stage 4 vs 5	Possible origin
*Group 1*
Isoprene	Gradual decrease until stage 3 (healthy vs stage 3: p = 0.006)	Gradual decrease until stage 3 (healthy vs stage 3: p = 0.006)	NS	NS	Cholesterol cycle
Acetone	Gradual decrease until stage 3 (healthy vs stage 3: p = 0.018)	Gradual decrease until stage 3 (healthy vs stage 3: p = 0.018)	0.025 (+)	NS	Acetone
*Group 2*
Ethylene glycol	NS	NS	0.018 (+)	NS	Sugar cycle metabolite or environmental toxin
Acetoin	NS	NS	0.005 (+)	NS	Sugar cycle metabolite or environmental (food additive or sweetener)
*Group 3a*
Styrene	0.042 (+)	NS	NS	NS	Environmental toxins
Toluene	0.028 (+)	NS	NS	NS
*Group 3b*
2,2,6-Trimethyl-octane	0.007 (+)	0.049 (-)	NS	NS	Potential CKD biomakers
2-Butatone	0.028 (+)	0.013 (-)	NS	NS
2,4-Dimethyl-heptane	Subsignificant increase	0.033 (-)	NS	NS
*Group 4*
2-Ethylpentane	NS	NS	NS	NS	No significance
Nonanal	NS	NS	NS	NS
O-Cymene	NS	NS	NS	NS
2-Methylhexane	NS	NS	NS	NS
2,2,3-Trimethyl-hexane	NS	NS	NS	NS
M-ethyl-methyl-benzene	NS	NS	NS	NS

Trends of the volatile organic compounds occurring in the breath samples of >75% of the test population. Increase or decrease in abundance are indicated by (+) and (−), respectively. The distinction between the patient subpopulations was studied via Wilcoxon/Kruskal–Wallis tests.
CKD: Chronic kidney disease; NS: Nonsignificant changes.

Comments

Commenting is limited to medical professionals. To comment please Log-in.

Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.