Evaluation of Methods for Detecting Endotoxin on Filters of Heating-, Ventilation- and Air-Conditioning Systems

Abstract

Endotoxin, a lipopolysaccharide component of the cell wall of gram negative bacteria, may play a major role in the development of the Sick Building Syndrome. Filter of heating-, ventilation- and air-conditioning systems can represent possible source of endotoxin contamination of indoor air. Measurements of endotoxin on filters cannot be compared between investigators because of different retrieving conditions. This study was inaugurated to determine the most effective method and their recovery rate for endotoxin retrieval. Determination of endotoxin was performed with a biological test (Limulus Amoebocyte Lysate). In this investigation the most common filter types of HVAC systems used in Germany were examined. These are cellulose-, glass- and synthetic fiber filters that were inoculated with four different types of endotoxins LPS in two different activities, Escherichia coli, Pseudomonas aeruginosa and dust out of an HVAC system). Filters were brand new to prevent pre-contamination because of impracticable detoxification of the material. Different contamination fluids were used to imitate most realistic (dust) circumstances on one and to perform most standardized (LPS) conditions on the other hand. To prepare endotoxin suspension by extracting the inoculated filters these were shaken or sonicated in a Tween 20 or Triethylamine buffer (TAP) solution. A total number of 360 samples were measured. In extensive preinvestigations optimal settings and conditions were drawn up. Detoxification with -radiation was checked, improvement of the test system, times of extraction and optimal concentration of solution media were worked out. Strict prevention of endotoxin contamination and a partly blinding were performed. For statistical evaluation descriptive data and data calculated with ?Generalized Linear Models? were shown. Because of inhomogeneous distribution of Pseudomonas aeruginosa endotoxin contaminated filters and its enormous influence on the results this strain was determined separately. When we analyzed data of all contaminations except Pseudomonas aeruginosa glass fiber filter showed a deviation from 100% recovery rate of 2.6%, cellulose fiber of 276.2% and synthetic fiber filter with 51.0%. Sonication revealed as best solution media with 18.9% difference ahead of shaking with 190.8%. TAP was supreme of Tween 20 with 53.3% difference and 160.6%, respectively. As expected LPS, it represents the most standardized endotoxin, showed best recovery of all examined endotoxins with a mean square fractional difference (MSFD) (2.1.8.2) of 70.1% for the activity of 200 EU and 48 % for 2,000 EU, respectively. Looking at the most standardized conditions with contamination of well defined endotoxin (LPS 2,000 EU) and a combination of the extraction method it was shown for cellulose fiber filter: Best extraction is performed with sonication in TAP with recovery of - 6.1% and an MSFD of 14.1%. In worst case if sonicated in Tween difference will present with -91.4% and an MSFD of 91.6%. Extraction of glass fiber filter is best if shaken in TAP (Figure 5 1) with MSFD of 19.2% and a difference of -16.7%. Almost as good as was the removal with sonication and TAP is with MSFD of 29.2% and an overestimation of 27.9%. Synthetic fiber filter revealed almost same results for all four combinations. Smallest difference for Tween and sonication with an overestimation of 5.5% is shown. MSFD for these combinations was between 20.3% and 24.1%. When calculating data with ?Generalized Linear Models? it revealed no significant differences for any constellation. This might be due to differences of sample conditions and their relatively small number of five samples per unique method. The extraction methods are slightly but not statistical significant (n.s.) different (p=0.323). Sonication showed a mean difference of 2 (CI95 ?370 to 374) compared with the TA and was better than shaking which showed a mean difference of 263 (CI95 ?109 to 635). Solution media (n.s.) (p=0.327). TAP showed a mean difference of 3 (CI95 ?369 to 375) from the TA and was more potent than Tween 20 that showed a mean difference of 262 (CI95 -110 to 634). Type of filter (n.s.) revealed a ?p? of 0.375. Glass fiber filters showed a mean difference of 0.5 (CI95 ?455 to 456) that was better than synthetic fiber filters that showed a mean difference of 2.6 (CI95 ?453 to 458) and was better than cellulose fiber filters which showed a mean difference of 394 (CI95 ?62 to 850). The contamination fluids reported a ?p? of 0.393 and were not statistical significant. LPS 200 showed a mean difference of 0.5 (CI95 ?525.8 to 526.8) and was not as close as LPS 2,000 that showed a mean difference of 0.2 (CI95 ?526.1 to 526.6) which was better than Dust that showed 1.0 for mean difference (CI95 ?525 to 527) and Escherichia coli which showed 528 (CI95 of 2 to1,054) for mean difference. Although using a sophisticated statistical analysis, precise preparation of sample contamination with well defined endotoxin solutions, sample extraction and accurate assay work with partly blinding it was not possible to find one significant, most reliable and precise method for endotoxin recovery on HVAC systems filter. There were major differences between filter material and types of endotoxin. Tendencies of better methods were observed and need to be confirmed in larger extent. Especially the recovery rate has to be determined prior to the investigation and should be labeled because of its major impact for the results. A preinvestigation study of the interesting filter type with LPS contamination and approximately 50 reiterations could attain the recovery rate. Most important differences can arise if the standard curve is not prepared with the same dilution media like used for endotoxin recovery. As a general recommendation it is mentioned to use endotoxin free material wherever possible, to check water for absence of endotoxin, to perform very precise work, to verify nonattendance of interference for every single probe, to use same lot number of test kits for all tests and to do the assay work immediately following the extraction. Especially the samples should be dried after sampling when immediately extraction and determination is not possible, because of uncertain recovery for suspended samples.

Endotoxin, eine Lipopolysaccharid-Komponente (LPS) der Zellwand gram-negativer Bakterien, kann eine Rolle in der Entwicklung des Sick Building Syndrome spielen. Filter von Klimaanlagen kommen als potentielle Quelle für Endotoxin- Kontamination der Innenluft in Frage. Die Messergebnisse zwischen Untersuchern sind nicht vergleichbar, weil keine standardisierte Methode vorhanden ist. In dieser Studie werden die effektivsten Methoden beschrieben zur Widerfindung von Endotoxinen auf Filtern von Klimaanlagen, mit den jeweiligen Widerfindungsraten. Die Bestimmung der Endotoxine erfolgte mit einem biologischen Test (Limulus Amoebocyte Lysate). Es wurden die am häufigsten verwendeten Filterarten in deutschen Klimaanlagen untersucht. Diese sind Zellulose-, Glass- und Synthesefaser Filter. Sie wurden mit vier verschiedenen Arten von Endotoxinen (Escherichia coli, Pseudomonas aeruginosa, gereinigtes Standard Endotoxin und Staub aus eine Klimaanlage) in zwei verschiedenen Konzentrationen beimpft. Das Filtermaterial wurde aus der Herstellung entnommen und vor Kontamination geschützt, da eine Detoxifikation nicht praktikabel ist. Unterschiedliche Kontaminationen wurden verwendet um möglichst realistische (Staub) und möglichst standardisierte (LPS) Versuchsbedingungen zu erzielen. Die beimpften Filter wurden in zwei Extraktionslösungen (Tween 20 oder Triethylamine buffer (TAP)) ausgeschüttelt oder mit einem Ultraschallbad behandelt. Eine Gesamtzahl von 360 Proben wurden gemessen. In umfangreichen Voruntersuchungen wurden die optimalen Untersuchungsbedingungen ermittelt. Detoxifikation mit Gamma-Stahlung wurde untersucht, Verbesserungen des Testsystems erarbeitet, sowie optimale Zeiten für die Extraktion bestimmt. Strikte Vermeidung der Kontamination mit Endotoxinen und eine teilweise Verblindung der Untersuchung wurden durchgeführt. Die statistische Auswertung erfolgte als beschreibender und berechneter Teil nach "Generalized Linear Models". Wegen inhomogener Verteilung der Messwerte für die Kontamination mit Pseudomonas aeruginosa Endotoxin erfolgte die statistische Auswertung für diesen Keim separat. Obwohl eine ausgeklügelte statistische Methode, präzise Präparation der Proben und die Kontamination mit gut definierten Endotoxin-Lösungen durchgeführt wurde, war es nicht möglich signifikante Unterschiede zwischen den Methoden zu erarbeiten. Es wurden größere Unterschiede zwischen den Filtermaterialien und Arten von Endotoxin festgestellt. Tendenzen für bessere Methoden wurden beobachtet und müssen in größeren Stichproben verifiziert werden. Vor einer Untersuchung müssen insbesondere die Wiederfindungsraten für das jeweilige Verfahren ermittelt und für die Vergleichbarkeit angegeben werden. Besondere Wichtigkeit hat die Anfertigung der Standardkurve, welche mit dem gleichen Lösungsmittel, wie die Extraktion, geschehen sollte. Als allgemeine Empfehlungen für die Endotoxinbestimmung auf Filtermaterialien lassen sich folgende Punkte anführen. Es sollte unter allen Umständen Endotoxin freies Material Verwendung finden. Das Wasser zur Verdünnung ist auf Endotoxine zu untersuchen. Störfaktoren für den biologischen Test müssen ausgeschlossen sein. Es sollten alle Bestimmungen mit der selben Charge des LAL-Reagenz durchgeführt werden.