Scalp hair is increasingly being used as a testing substrate for toxins and monitoring treatment adherence. The biochemistry of human hair is assumed to be similar; however, a recent study reported higher concentration of lipids in African hair. The effect of hair curvature, if any, on drug incorporation (e.g. lipid soluble drugs) is unknown. Racial description of hair morphology is unscientific. A geometric classification of hair into 8 groups (I-straight to VIII-tightly curly) was recently proposed, however its reliability has not been confirmed. The aim of this thesis was to investigate the reliability of the geometric classification (and to assess whether it could be improved) and investigate relationships between morphology and other hair characteristics. Virgin hair was collected from 128 volunteers using a standardized protocol. Geometric measurements of hair using published templates were conducted for classification. Reliability was assessed using Kappa statistics. Characteristics assessed included mechanical properties (miniature tensile tester), growth rate and hair density (TrichoScan® trichogram), biochemistry (Vanillin assay for lipids and Fourier Transform Infrared adsorption) and imaging (Electron and Fluorescent Light Microscopy). Inter-observer agreement was poor for 8-groups (k=0.418) but improved for 6-groups (k=0.671). The intra-observer agreement also improved [ranges: k=0.444 to 0.648 (8-groups) and k=0.599 to 0.836 (6-groups)]. The yield strength of all hair groups was higher than reported for racially grouped samples. Curly hair groups had lower growth rates and tensile strengths. The TrichoScan based growth rate was for fastest for the straightest (0.72±0.3 cm/month) and slowest for the curliest (0.39 ± 0.2 cm/month) hair. No correlation with biochemistry was detected for either the 8 or 6-group classification, although a trend toward higher absorption of lipid (C-H) bands was noted for curly hair. A supervised statistical approach applied to 4 hair groups using the FTIR data improved classification success to 79% (range: 69% - 88%), which needs confirmation but would be more objective than using race for hair testing in Medicine and Forensic Science. This thesis supports a geometric classification with fewer groups (6, based kappa statistics and 4, based on biochemistry); it is also the first to report correlations between hair geometry, biochemistry and physical properties.

Background Acute Myocardial infarction (MI) or heart attack is a leading cause of death worldwide. Since an MI is a stressful life event, plasma cortisol levels are expected to increase significantly from the baseline. Cortisol and dehydroepiandrosterone (DHEA) as products of the hypothalamicpituitary-adrenal (HPA)-axis have been used to diagnose endocrine disorders in serum, urine and saliva. However, these body fluids reflect short-term assessments. Hair is a promising alternative and offers several advantages over serum e.g. hair collection is painless, provides a longer detection window (days to months depending on length) and is easy to store. Recent studies confirm good correlation between hair cortisol levels and validated stress questionnaires (i.e., hair cortisol is a biomarker of stress). It is unclear whether stress is a result or cause of acute MI (i.e., chicken or egg). The primary aim of this study was to determine whether hair cortisol is higher in patients with acute MI versus healthy controls and whether the higher levels pre-date acute MI (i.e. is hair cortisol a reliable biomarker of chronic stress that predates acute MI). Methods This pilot study aimed to include 25 age-matched controls and acute MI cases, both 25 STelevation myocardial infarctions (STEMIs) and 25 non-ST-elevation myocardial infarctions (NSTEMIs). Complete versus partial coronary artery occlusion is associated with STEMIs and NSTEMIs respectively. Cortisol levels were measured in 3 cm segments of hair representing 9-12 months of growth (which would be before the heart attack in acute MI cases). The proximal 3 cm closest to the scalp, is estimated to represent 3 months before heart attack. The samples were prepared and analyzed for cortisol concentrations using an enzyme-linked immunoassay kit. In addition, at least three strands of hair from each participant were mounted on a glass slide and subjected to Fourier-Transform Infrared (FTIR) spectroscopy. For cortisol concentrations two-sample Wilcoxon rank-sum (Mann-Whitney) was used for within an individual comparisons and Kruskal-Wallis test was used for within group comparisons (P<0.05 considered significant) For FTIR data Principal Component Analysis (PCA), Orthogonal Projections to Latent Structures Discriminant Analysis (OPLS-DA) and Partial Least Squares (PLS) models were used to compare groups. Results Of the intended sample size of 75, 64 participants (48 acute MI cases and 16 healthy controls) were recruited. The total hair cortisol concentration was significantly higher in acute MI cases than controls [median 23.66 (3.73-209.18) vs. 3.32 (0.37-11.24) pg/mg], (p<0.001), and higher in the hair of STEMI versus NSTEMI cases[35.18 (8.15-209.180 vs. 17.24 (3.73-148.22) pg/mg], (p< 0.01). Further, cortisol levels were similar in proximal versus distal hair segments (within an individual) of STEMIs (p< 0.05) but significantly different in those of NSTEMIs [29.64 (5.87- 148.42) vs 9.94 (5.15-32.95)], (p=0.002). In view of the fact that all 16 controls included were female, the analysis was repeated to include only female study participants[16 controls and 27 cases (10 STEMIs and 17 NSTEMIs)]. The total hair cortisol concentration was also significantly higher in female acute MI cases than controls [median 21.59 (3.73-209.18) vs. 3.32 (0.37-11.24)], (p< 0.001), as well as higher in the hair of STEMI vs. NSTEMI cases [median 37.21 (8.15-209.18) vs. 14.11 (3.73- 148.22pg/mg], (p< 0.05). Further, cortisol levels were similar in proximal versus distal hair segments of STEMIs (p< 0.05) but significantly different in those of NSTEMIs [median 15.26 (5.91-100.41) vs. 9.94 (3.73-37.57)], (p< 0.01). Controls had the lowest cortisol levels and there was no significant difference between proximal and distal hair within the individuals [median 3.18(1.43-7.09) vs. 2.30 (0.68-4.62) pg/mg], (p> 0.05). For FTIR data, OPLS-DA showed a separation between acute MI cases and controls, as well as a separation between STEMIs and NSTEMIs. Further, there was also good correlation between total cortisol concentrations and FTIR spectral data (correlation coefficient = 0.85). Study limitations Although not all intended participants were recruited during the study period, repeating the analysis for only female participants did not alter the findings. Discussion and Conclusions Hair cortisol levels were higher in acute MI cases than controls and significantly higher in STEMIs versus NSTEMIs. Further, FTIR data separated acute MI cases from controls and correlated with cortisol concentrations. Patients with Cushing Disease (CD) are reported to have a higher prevalence of thromboembolic events compared to Cushing Syndrome from adrenal sources, this is thought to be a result of high cortisol levels associated with CD. In this study, STEMIs had the highest cortisol levels which not only predated the acute MI but were unchanged in hair segments correlating with the preceding 9 months (i.e., were chronically consistently high). Interestingly, high cortisol levels also predated the acute MI but more than doubled in the proximal 3 cm hair segments (correlating with 3 months) before the heart attack in NSTEMIs. This study suggests that based on hair cortisol concentrations as a biomarker, chronic constantly high stress predates STEMIs. Although also high in NSTEMIs it is lower than in STEMIs and more than doubles before the onset of an acute MI. The effects of chronic very high cortisol levels on the thromboembolic pathway may be to cause complete (versus partial) coronary artery occlusion in STEMIs. These results require validation in larger studies.

Background: Diabetes mellitus is a major public health problem resulting in about 5 million deaths per year. This metabolic disorder is characterized by hyperglycaemia, which results in debilitating and life-threatening complications. It is, therefore, vital for diabetics to monitor and control their blood glucose levels in order to keep them below 7mmol/L while fasting and below 9mmol/L after meals. Chronic estimates of glucose control of 8-12 weeks are obtained using glycated haemoglobin A1 (HbA1c). Non-invasive, less expensive methods of monitoring long term glycaemic control may be useful. Since scalp hair consists of about 80% protein, which is subject to non-enzymatic glycation, and growing hair has a rich blood supply exposing it to free glucose, it is likely that hair can be used as an alternative substrate for monitoring chronic hyperglycaemia. Subjects and Methods: Scalp hair and a blood samples (for HbA1c) were collected from 46 diabetic and 46 healthy control subjects. There were 26 diabetic adults (30-70 years), recruited from the outpatient clinic at Groote Schuur hospital and 20 children (7-18 years) recruited from the diabetic clinic at the Red Cross children's hospital. There were 29 healthy control adults (26-65 years) and 17 children (7- 17 years) recruited from the Groote Schuur and Red Cross hospitals respectively. History of chemical hair treatment was recorded for each participant. Hair samples were washed using 1% sodium dodecyl sulphate and analysed using Fourier transform infrared- attenuated total reflection (ATR-FTIR) spectroscopy. Spectra were analysed using statistical software (SIMCA, Umetrics) to determine whether the hair of diabetics was distinguishable from hair of healthy controls as well as whether spectra correlated with HbA1c levels of participants. Hair amino acid concentrations were also analysed as it is known that circulating amino acid concentrations are altered in people with diabetes. Results and discussion: The Orthogonal Projections to Latent Structures Discriminant Analysis (OPLS-DA) models between spectra obtained from hair of diabetic participants and spectraobtained from control hair show good separation and predictive ability. When ATR-FTIR spectra were analysed in four groups: children with natural hair, adults with natural hair, adults with dyed hair and adults with relaxed hair, the models yielded predictive abilities (Q2(cum)) ranging from 0.753 to 0.85 and classification abilities ranging from 97.97 to 100%. This demonstrated excellent separation and predictive ability for controls vs. diabetics. Partial least squares (PLS) revealed a good correlation between hair FTIR spectra and participant HbA1c levels (R2 ranging between 0.8067 and 0.9296). These results demonstrated the possibility to use ATR-FTIR alongside multivariate data analysis to detect hyperglycaemia and monitor blood glucose levels via prediction of HbA1c levels from the hair spectra. Amino acid analysis supported the OPLS-DA classifications, as the largest differences were seen between age and chemically treated hair groups. Amino acid results reinforced the necessity to classify spectra into groups in order to distinguish between hair spectra from diabetics or controls, as well as to predict HbA1c. Twelve amino acids (Asp, Glu, Pro, Gly, Met, Ile, His, Lys, Arg, Amm, Cys, Leu) were significantly different between hair from adults and children, eight amino acids (Ala, Cys, Val, Met, Lys, Amm, Ser, Tyr) were statistically significantly different between natural and dyed hair and only up to four amino acids (Gly, Val, Met, Ile in children or Val, Phe, Amm, Arg in adults) were significantly different between diabetic and control groups. Conclusion: There is a need for non-invasive means of monitoring chronic hyperglycaemia. This study demonstrated the ability to distinguish between the hair of diabetics and controls as well as the ability to predict HbA1c levels from hair using ATR-FTIR. However, factors such as age and chemical treatment, which affect the chemical properties of hair, like amino acid levels, should be considered first. This would lead to promising prospects for long term blood glucose monitoring, due to the ability to estimate hair growth rate, and greater insights into the timing and development of diabetic complications. ATR-FTIR relatively simple to use, requires minimal sample preparation and does not require the use of expensive consumables. This technology could, potentially, be adapted into a primary health point of care or home screening or monitoring device for long-term hyperglycaemia, which would assist in early detection and preventing the progression of debilitating complications.

Excessive scarring due to fibrosis from aberrant wound healing can lead to conditions such as Keloids or Folliculitis Keloidalis Nuchae (FKN). These fibroproliferative growths pose therapeutic challenges due to their complex aetiology that has been linked to multiple genetic and environmental factors, with frequent reoccurrence following therapy. Owing to reports on an increase in ATP and Fibroblast Activation Protein-1α production in keloids, the aim of this study was to investigate whether the disease phenotypes were linked to bioenergetic changes at a cellular level in these two conditions. Patient-derived tissue biopsies were used for fibroblast cell culture models, in which cell analysis was carried out to assess phenotype and different parameters of bioenergetic cellular behaviour. In addition to FKN and the intra- and peri-lesional Keloid patient fibroblasts, normal skin and non-hypertrophic (normo-trophic) scar fibroblasts were used as negative controls. The results show statistically significant and variable growth dynamics with increased proliferation and migration in keloid fibroblasts, while FKN fibroblasts showed a statistically significant increase in proliferation but had a similar migration profile to controls. The results further show that there is a statistically significant metabolic switch towards aerobic glycolysis in the fibroblasts from the disease conditions. During functional measurement of mitochondrial parameters, an increase in oxidative phosphorylation was exhibited in the disease conditions indicating their mitochondria were still functional. An increase in basal glycolysis with a concomitant increase in the cellular maximum glycolytic capacity was also demonstrated. Furthermore, protein analysis showed an upregulation in the expression of Fibroblast Activation Protein-1α in fibroblasts from both disease conditions This study begins to give novel insight into the bioenergetics of normal scars and scarring conditions such as FKN and, adds to the knowledge on the heterogeneity of fibroblasts derived from specific lesional sites within Keloids. These findings suggest that Keloids and FKN have a switch to a metabolic phenotype of aerobic glycolysis. This increase in glycolytic flux potentially proposes glycolytic inhibitors as a mechanistic basis for the treatment of these conditions.