Thesis (MMed)--Stellenbosch University, 2014. / ENGLISH ABSTRACT: Background: Sputum is a non-invasive, economic investigation whereby bronchogenic carcinoma can be identified. Manual cytological screening is labour intensive, time-consuming and requires a continuous high level of alertness. Automation has recently been successfully introduced in gynaecological cytology. Since sputum samples are similar to cervical smears, the question arises as to whether they are also suitable for automated screening. Objective: This study presented with various objectives: 1) To test automated sputum screening using the BD FocalPoint™ Slide Profiler (FP) and compare with manual sputum screening. 2) To determine the sensitivity and specificity of sputum in identification of bronchogenic carcinoma. 3) To ascertain if any clinical, radiological or bronchoscopy findings would be predictors for bronchogenic carcinoma. 4) To determine the significance of adequacy. Method: Sputum samples were collected prospectively from patients attending the Division of Pulmonology at Tygerberg hospital for a transbronchial fine needle aspiration biopsy (TBNA) or a transthoracic fine needle aspiration biopsy (TTNA) for the period from 2010 to 2012. A pre-bronchoscopy sputum was collected and submitted for processing. Stained slides were put through the FP for automated screening. After slides were qualified, sputum slides were put back in the routine screening pool. Correlation was done using the TBNA/TTNA result as the standard to evaluate the sputum results. Results: 108 sputum samples were included in this study. Of the 84.3% malignant (n=91) and 15.7% benign (n=17) cases confirmed with a diagnostic procedure, sputum cytology had a sensitivity of 38.5% (35/91 malignant cases), and a specificity of 100% (17/17 benign cases). Automated screening had a better sensitivity of 94.3% (33/35 positive sputum cases), while manual screening showed a sensitivity of 74.3% (26/35 positive sputum cases) when compared to the final sputum result. Individual parameters with a significant association with positive sputum included the presence of an endobronchial tumour, partial airway obstruction / stenosis, round mass, spiculated mass (negative association), loss of weight (negative association) and squamous cell carcinoma as the histological subtype. Adequacy was not as significant as hypothesised since 85.3% of true positive sputum, but also 65.5% of false negative sputum, had large numbers of alveolar macrophages present. Conclusion: Sputum cytology remains an important part of the screening programme for bronchogenic carcinoma in the public health sector of South Africa. Results confirm that sputum cytology is very specific, and automated screening improves sensitivity. Automated screening proved to be more time efficient, resulting in 83.1% reduction (p<0.0001) in the screening time spent per case by a cytotechnologist. Results confirm that the quantity of alveolar macrophages is not directly proprtional to pathology representation. Positive sputum results did however improve with sputum adequacy, but had no significant association. Recommendations from this study include adopting automated sputum screening. / AFRIKAANSE OPSOMMING: Agtergrond: Die verkryging van ‘n sputummonster is ‘n nie-indringende, ekonomiese ondersoek waardeur bronguskarsinoom identifiseer kan word. Nie-geoutomatiseerde sitologiese ondersoek is arbeidsintensief, tydrowend en vereis ‘n deurlopende hoë vlak van konsentrasie en fokus. Outomatisering is onlangs suksesvol geïmplementeer in ginekologiese sitologie-ondersoeke. Aangesien sputummonsters soortgelyk aan servikale monsters is, het die vraag ontstaan of sputummonsters ook geskik sou wees vir geoutomatiseerde sifting. Doelwit: Hierdie studie het verskeie doelwitte gehad: 1) Om geoutomatiseerde sifting van sputummonsters te toets deur gebruik te maak van BD Focal Point ™ Slide Profiler (FP), en te vergelyk met nie-geoutomatiseerde sputum sifting. 2) Om die sensitiwiteit en spesifisiteit van sputum in die identifikasie van bronguskarsinoom te bepaal. 3) Om vas te stel of enige kliniese, radiologiese of brongoskopiese bevindings bronguskarsinoom sou kon voorspel. 4) Om die belang van ‘n verteenwoordigende monster te bepaal. Metode: ‘n Prospektiewe studie van die pasiënte wat die Divisie van Pulmonologie by Tygerberg Hospitaal vir transbrongiale nodale aspirasie (TBNA) of ‘n transtorakale aspirasie (TTNA) vanaf Julie 2010 tot Mei 2012 bygewoon het, is gedoen. ‘n Prebrongoskopiese sputum is geneem en gestuur vir prosessering. Die gekleurde skuifies is deur die FP gestuur vir geoutomatiseerde ondersoek. Indien die sputumskuifies gekwalifiseer het vir geoutomatiseerde sifting, is hulle in die groep vir ondersoek ingesluit. ‘n Korrelasiestudie, om die sputumresultate te evalueer, is uitgevoer deur die TBNA/TTNA bevindings as standaard te gebruik. Resultate: Vir hierdie studie is 108 sputummonsters ingesluit. Vanuit die 84.3% maligne (n=91) en 15.7% benigne (n=17) gevalle, bevestig deur ‘n diagnostiese prosedure, het sputumsitologie ‘n sensitiwiteit van 38.5% (35/91 maligne gevalle) en ‘n spesifisiteit van 100.0% (17/17 benigne gevalle), getoon. Geoutomatiseerde sifting het ‘n beter sensitiwiteit met 94.3% (33/35 maligne gevalle), terwyl nie-geoutomatiseerde (ondersoek) ‘n sensitiwiteit van 74.3% (26/35 maligne gevalle) wanneer met die finale resultaat vergelyk, gevind. Individuele parameters met ‘n betekenisvolle assosiasie het die teenwoordigheid van ‘n endobrongiale tumor, gedeeltelike lugwegobstruksie / stenose, ronde massa, ‘n spekuleerde massa (negatiewe assosiasie), gewigsverlies (negatiewe assosiasie) en plaveiselkarsinoom as die histologiese subtipe, ingesluit. Geskiktheid van die monster was nie so betekenisvol as wat in die hipotese gestel is nie: aangesien 85.3% van ware positief gediagnoseerde sputummonsters, maar ook 65.5% van die vals negatiewe sputummonsters, groot hoeveelhede alveolêre makrofae ingesluit het. Gevolgtrekking: Sputumsitologie bly steeds ‘n belangrike deel van die siftingsprogram vir bronguskarsinoom in die openbare gesondheidssektor in Suid-Afrika. Resultate van hierdie studie bevestig dat sputumsitologie baie spesifiek is en dat geoutomatiseerde sifting die sensitiwiteit verbeter. Ge-outomatiseerde sifting het bewys dat dit meer tydsbesparend is, met ‘n 83.1% vermindering (p<0.0001) in die siftingstyd wat deur een sitotegnoloog per geval bestee word. Resultate het bevestig dat die hoeveelheid alveolêre makrofae nie direk proporsioneel verwant is tot die patologie nie. Hoe meer verteenwoordigend die sputummonster was, hoe groter was die kanse om ‘n akkurate positiewe diagnose te maak. Die assosiasie van die geskiktheid van die sputummonster en die positiewe resultate het egter nie ‘n statisties betekenisvolle resultaat getoon nie. Aanbevelings vir hierdie studie sluit in die aanwending van geoutomatiseerde sputumondersoeke.