Background: Iron deficiency (ID) and hemoglobinopathies are highly prevalent in Southeast Asia. Accurate estimation of reference intervals (RIs) for red cell parameters is complicated by the need to exclude individuals with these conditions from the reference population. Indirect RI estimations using machine learning could help overcome these challenges. Methods: We developed a binary classification model using eXtreme Gradient Boosting (XGB) to distinguish normal individuals from those with ID, hemoglobinopathies, or other anemias. The model was trained on an annotated dataset comprising 5,520 complete blood count (CBC) results and validated with a holdout dataset of 2,367 CBC results. An independent dataset of 64,100 CBC results was used to identify individuals predicted to be normal, from which RIs were estimated using the refineR algorithm. Results: The XGB model achieved an area under the ROC of 0.97 (95% confidence interval: 0.96–0.97) for distinguishing between individuals with normal versus abnormal values. Among individuals within the independent dataset, 40,300 (62.9%) were predicted to be normal. The refineR-based reference limits (RLs) derived from this subset approximated those obtained through a direct approach. Improvements in the accuracy of indirect RL estimates were most evident for hematocrit, hemoglobin, and red cell concentrations. Conclusions Combining XGB with refineR to indirectly derive RIs for red cell parameters improved the accuracy and yielded results comparable with those of directly derived RIs. A further benefit was the capacity to generate sex- and age-specific ranges, which has remained difficult to achieve through direct approaches.

Introduction: Non-diabetic hypoglycemia in older adults warrants careful evaluation across insulin-mediated and noninsulin-mediated causes. We present a challenging case of insulin autoimmune syndrome (IAS; Hirata’s syndrome). Case: An 85-year-old female presented with severe hypoglycemia (capillary blood glucose [CBG] 1.8 mmol/L) with reduced consciousness, preceded by 2 months of recurrent dizziness relieved by food intake. Appetite and weight were stable. Past medical history included stage 3 chronic kidney disease and osteoporosis, but no diabetes. Medication review revealed a recent 2-week course of traditional supplement, long-term Neurobion®, but no agents with recognized hypoglycemic potential. Physical examination showed a moderately built elderly female with a body mass index of 24.3 kg/m² and no Cushingoid features. Recurrent hypoglycemia (CBG nadir 1.5 mmol/L) occurred in both fasting and postprandial states, fulfilling Whipple’s triad. Baseline investigations showed an estimated glomerular filtration rate of 42 mL/min/1.73 m², AM cortisol of 700 mmol/L, and unremarkable liver and thyroid function tests. During a hypoglycemic episode (CBG 2.3 mmol/L), plasma insulin and C-peptide were inappropriately raised at 203.6 mU/L (reference interval [RI]: 3.0– 25.0) and 33 ng/mL (RI: 0.9–7.1), respectively, confirming endogenous hyperinsulinemic hypoglycemia. Polyethylene glycol precipitation showed 21% insulin recovery, raising suspicion for an autoimmune cause. Elevated insulin autoantibodies (175 AU/mL; RI <20) confirmed IAS. Computed tomography of the abdomen and endoscopic ultrasound excluded a pancreatic neuro-endocrine tumor. Nutritional management comprising frequent low glycemic index feeds and uncooked cornstarch was commenced. Diazoxide 100 mg TDS caused fluid overload and severe hyponatremia, while hypoglycemia persisted at lower doses, necessitating discontinuation. Subcutaneous octreotide 100 mg QID was required to control hypoglycemia. Prednisolone 30 mg BD improved glycemic stability. Insulin autoantibodies titer remained 175 AU/mL at 2 weeks; reassessment was conducted at 4 weeks with consideration for biologics if persistent. : doses, necessitating discontinuation. Subcutaneous octreotide 100 mg QID was required to control hypoglycemia. Prednisolone 30 mg BD improved glycemic stability. Insulin autoantibodies titer remained 175 AU/mL at 2 weeks; reassessment was conducted at 4 weeks with consideration for biologics if persistent. Conclusion Endogenous hyperinsulinemic hypoglycemia, after excluding insulinoma, should raise suspicion for IAS. Management includes removing triggers, supportive care, and immunomodulatory therapy in severe cases.

Introduction: Thyroid hormone resistance syndrome (THR) is a disorder characterized by reduced responsiveness of target tissues to thyroid hormones with non-suppressed thyroidstimulating hormone (TSH) despite elevated free thyroxine 4 (FT4). Diagnosing and managing THR in athyreotic patients, however, is challenging; indeed, TSH levels post thyroidectomy for Differentiated Thyroid Carcinoma in patients with THR have been reported to reach levels as high as 112.59 mIU/L despite high-dose thyroxine-suppressive therapy. This case highlights the complexities in diagnosing Resistance to Thyroid Hormone (RTH) post-thyroidectomy. Case: A 63-year-old female with end-stage renal failure (ESRF) on hemodialysis and prior parathyroidectomy for tertiary hyperparathyroidism underwent total thyroidectomy in 1997 for presumed benign goiter. Following surgery, she demonstrated persistently elevated TSH, ranging from 93.2 to 670.4 mIU/L (0.55–4.78 mIU/L), with normal to mildly elevated FT4 levels, ranging from 17 to 35 pmol/L (11.5–22.7 pmol/L), while on thyroxine replacement doses as low as 0.86 ug/kg. Discordant thyroid function tests (TFTs) were consistent across different assay platforms. Polyethylene glycol precipitation excluded macro-TSH interference. Intermittent levothyroxine increments suppressed TSH but led to thyrotoxic symptoms, including weight loss, heat intolerance, insomnia, and fragility fracture. Uncontrasted pituitary magnetic resonance imaging (risk of nephrogenic systemic fibrosis with gadolinium in ESRF) showed a small pituitary gland without adenoma, excluding TSH-oma. T3 suppression test was relatively contraindicated due to her advanced age and co-morbidities. There was no family history of thyroid disorder; the patient’s only child had a normal TFT, and her parents/siblings could not be tested. A working diagnosis of RTHβ was made. She declined genetic testing for RTH. She is currently on levothyroxine 50 mcg OD (1.07 ug/kg) with FT4 16.1 pmol/L and TSH 562 mIU/L with no symptoms/signs of hypo or hyperthyroidism. Conclusion This case highlights the diagnostic challenges in managing possible RTHβ post-total thyroidectomy. Treatment should be guided by clinical status and target of mid-to-normal FT4 rather than TSH to avoid iatrogenic thyrotoxicosis.
Thyroidectomy ; Thyroid Hormones

INTRODUCTION: The diagnosis of Wilson disease (WD) is plagued by biochemical and clinical uncertainties. Thus, calculated parameters have been proposed. This study aimed to: (a) compare the diagnostic values of non-caeruloplasmin copper (NCC), NCC percentage (NCC%), copper-caeruloplasmin ratio (CCR) and adjusted copper in WD; and (b) derive and evaluate a discriminant function in diagnosing WD. METHODS: A total of 213 subjects across all ages who were investigated for WD were recruited. WD was confirmed in 55 patients, and the rest were WD free. Based on serum copper and caeruloplasmin values, NCC, NCC%, CCR and adjusted copper were calculated for each subject. A function was derived using discriminant analysis, and the cut-off value was determined through receiver operating characteristic analysis. Classification accuracy was found by cross-tabulation. RESULTS: Caeruloplasmin, total copper, NCC, NCC%, CCR, adjusted copper and discriminant function were significantly lower in WD compared to non-WD. Discriminant function showed the best diagnostic specificity (99.4%), sensitivity (98.2%) and classification accuracy (99.1%). Caeruloplasmin levels <0.14 g/L showed higher accuracy than the recommended 0.20 g/L cut-off value (97.7% vs. 87.8%). Similarly, molar NCC below the European cut-off of 1.6 umol/L showed higher accuracy than the American cut-off of 3.9 umol/L (80.3% vs. 59.6%) (P < 0.001). NCC%, mass NCC, CCR and adjusted copper showed poorer performances. CONCLUSION Discriminant function differentiates WD from non-WD with excellent specificity, sensitivity and accuracy. Performance of serum caeruloplasmin <0.14 g/L was better than that of <0.20 g/L. NCC, NCC%, CCR and adjusted copper are not helpful in diagnosing WD.
Humans ; Hepatolenticular Degeneration/diagnosis* ; Copper/analysis* ; Ceruloplasmin/metabolism* ; Repressor Proteins

Introduction: The Malaysian Association of Clinical Biochemists (MACB) established a Task Force for Chronic Kidney Disease. A survey was undertaken by the Task Force on the reporting of estimated glomerular filtration rate (eGFR) and urine albumin by hospital laboratories in Malaysia in both the government and private sectors. Materials and Methods: An e-mail invitation to participate in an online survey was sent to hospital laboratories in Malaysia (n=140). Questions regarding methods for measuring creatinine, equations for calculating eGFR, eGFR reporting, the terminology used in reporting urine albumin, types of samples and the cut-off values used for normal albuminuria. Results: A total of 42/140 (30%) laboratories answered the questionnaire. The prevalent method used for serum creatinine measurement was the Jaffé method (88.1%) traceable to isotope-dilution mass spectrometry. eGFR was reported along with serum creatinine by 61.9% of laboratories while 33.3% of laboratories report eGFR on request. The formula used for eGFR reporting was mainly MDRD (64.3%) and results were reported as exact numbers even when the eGFR was >60 ml/min/1.73m2 . The term microalbumin is still used by 83.3% of laboratories. There is a large heterogeneity among the labs regarding the type of sample recommended for measuring urine albumin, reference interval and reporting units. Conclusion: It is evident that the laboratory assessment of chronic kidney disease in Malaysia is not standardised. It is essential to provide a national framework for standardised reporting of eGFR and urine albumin. Recommendations developed by the MACB CKD Task Force, if adopted by all laboratories, will lead to a reduction in this variability.

Introduction: Hyperandrogenism remains as one of the key features in Polycystic Ovarian Syndrome (PCOS) and can be assessed clinically or determined by biochemical assays. Hirsutism is the most common clinical manifestation of hyperandrogenism. The clinical assessment is subjected to wide variability due to poor interobserver agreement and multiple population factors such as ethnic variation, cosmetic procedures and genetic trait. The difficulty in resolving the androgen excess biochemically is due to a lack of consensus as to which serum androgen should be measured for the diagnosis of PCOS. The aim of the study was to compare and establish the diagnostic cut off value for different androgen biomarker for the diagnosis of PCOS. Materials and Methods: A total of 312 patients classified to PCOS (n = 164) and non PCOS (n = 148) cohorts were selected from the Laboratory Information System (LIS) based on serum total testosterone (TT) and sex hormone binding globulin (SHBG) from the period of 1st April 2015 to 31st March 2016. PCOS was diagnosed based on Rotterdam criteria. Clinical hyperandrogenism and ultrasound polycystic ovarian morphology were obtained from the clinical records. The other relevant biochemical results such as serum luteinizing hormone (LH), follicle stimulating hormone (FSH) and albumin were also obtained from LIS. Free androgen index (FAI), calculated free testosterone (cFT) and calculated bioavailable testosterone (cBT) were calculated for these patients. Receiver Operating Characteristic (ROC) curve analysis were performed for serum TT, SHBG, FAI, cFT, cBT and LH: FSH ratio to determine the best marker to diagnose PCOS. Results: All the androgen parameters (except SHBG) were significantly higher in PCOS patients than in control (p<0.0001). The highest area under curve (AUC) curve was found for cBT followed by cFT and FAI. TT and LH: FSH ratio recorded a lower AUC and the lowest AUC was seen for SHBG. cBT at a cut off value of 0.86 nmol/L had the highest specificity, 83% and positive likelihood ratio (LR) at 3.79. This is followed by FAI at a cut off value of 7.1% with specificity at 82% and cFT at a cut off value of 0.8 pmol/L with specificity at 80%. All three calculated androgen indices (FAI, cFT and cBT) showed good correlation with each other. Furthermore, cFT, FAI and calculated BT were shown to be more specific with higher positive likelihood ratio than measured androgen markers. Conclusions: Based on our study, the calculated testosterone indices such as FAI, cBT and cFT are useful markers to distinguish PCOS from non-PCOS. Owing to ease of calculation, FAI can be incorporated in LIS and can be reported with TT and SHBG. This will be helpful for clinician to diagnose hyperandrogenism in PCOS.

Hypophosphataemia is a metabolic disorder that is commonly encountered in critically ill patients.Phosphate has many roles in physiological functions, thus the depletion of serum phosphate could leadto impairment in multiple organ systems, which include the respiratory, cardiovascular, neurologicaland muscular systems and haematological and metabolic functions. Hypophosphataemia is defined asplasma phosphate level below 0.80 mmol per litre (mmol/L) and can be further divided into subgroupsof mild (plasma phosphate of 0.66 to 0.79 mmol/L), moderate (plasma phosphate of 0.32 to 0.65mmol/L) and severe (plasma phosphate of less than 0.32 mmol/L). The causes of hypophosphataemiainclude inadequate phosphate intake, decreased intestinal absorption, gastrointestinal or renal phosphateloss, and redistribution of phosphate into cells. Symptomatic hypophosphataemia associated withhaematological malignancies has been reported infrequently. We report here a case of asymptomaticsevere hypophosphataemia in a child with acute T-cell lymphoblastic leukaemia.A 14-year-old Chinese boy was diagnosed to have acute T cell lymphoblastic leukaemia (ALL).His serum biochemistry results were normal except inorganic phosphate and lactate dehydrogenaselevels. The serum inorganic phosphate level was 0.1mmol/L and the level was low on repeatedanalysis. The child had no symptoms related to low phosphate levels. The possible causes of lowphosphate were ruled out and urine Tmp/GFR was normal. Chemotherapy regime was started andthe serum phosphate levels started to increase. Hypophosphataemia in leukaemia was attributed toshift of phosphorus into leukemic cells and excessive cellular phosphate consumption by rapidlyproliferating cells. Several reports of symptomatic hypophosphataemia in myelogenous andlymphoblastic leukaemia in adults have been reported. To our knowledge this is the first case ofsevere asymptomatic hypophosphataemia in a child with ALL.

Introduction: Prolactin (PRL) exists in different forms in human serum. The predominant form is monomeric PRL (molecular mass 23 kDa) with smaller amounts of big PRL (molecular mass 50–60 kDa) and at times macroprolactin (molecular mass 150–170 kDa). Macroprolactin, generally considered to be biologically inactive, accounts for the major part of prolactin in some patients. Different immunoassays for prolactin differ in reactivity with this macromolecular complex. Aim: The present study was undertaken to assess the incidence of macroprolactinaemia in our cohort of hyperprolactinemic patients. Method: 204 samples with hyperprolactinemia were evaluated for macroprolactinemia by polyethylene glycol (PEG) precipitation and gel fi ltration chromatography (GFC). Recoveries <60% after PEG precipitation were considered to have macroprolactinaemia. Results: A total of 43 (21%) of these patients had less than 60% recovery after PEG precipitation. GFC confi rmed that in seven of these patients macroprolactin was the major part of the prolactin. Recoveries were < 40% PEG precipitation in these samples. Combined macro and hyperprolactinemia was observed in two samples and the recovery after PEG precipitation was >40% but 50%. The incidence of macroprolactinemia in our cohort of hyperprolactinaemic patients was noted to be 4.4%. Conclusion: Macroprolactin is a signifi cant cause of misdiagnosis, unnecessary investigation, and inappropriate treatment and hence it is useful to screen all patients with high PRL levels with PEG precipitation and to apply GFC to samples with recoveries <50%.

Aim: Autoantibodies against cyclic citrullinated peptide (anti-CCP) are considered to be a sensitive and specifi c marker for rheumatoid arthritis (RA). This study evaluated the diagnostic and analytical performances of the automated anti-CCP assay. Materials and Method: Sera from 80 patients with established RA, 65 from other rheumatic diseases (non-RA) and 55 from healthy controls were studied using second generation anti-CCP. Rheumatoid factor (RF) was also assayed in each sample, and the results were compared to the anti-CCP fi ndings. Serum pools were used to determine the precision and linearity. Results: At a cut-off of 7.4 U/ml for anti-CCP, the sensitivity and specifi city for RA were 65% and 98% respectively. RF had a sensitivity of 58% and a lower specifi city of 93 % than anti-CCP. Conclusion: The high specifi city of the assay suggests that anti-CCP is useful in the diagnosis of rheumatoid arthritis and in our cohort of study population anti-CCP exhibits a better diagnostic value than RF. A considerable proportion (28%) of RF-negative RA patients were anti-CCP positive. Based on analytical performance of the assay, we conclude that full automation and high throughput features of AxSYM makes it an ideal platform for routine testing of anti-CCP.