Observing the cyclical changes in cervical epithelium using infrared microspectroscopy

Observing the cyclical changes in cervical epithelium using infrared microspectroscopy
of 9
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  Observing the cyclical changes in cervical epitheliumusing infrared microspectroscopy Melissa J. Romeo, Bayden R. Wood, Don McNaughton  Centre for Biospectroscopy and School of Chemistry, P.O. Box 23, Monash University, Victoria 3800, Australia Abstract We investigated the hormonal influences on cervical cells using infrared microspectroscopy and found that there wereobservable spectral changes occurring throughout the cycle. The main differences were seen in the glycogen region (1200–1000 cm  1 ) and the greatest cyclical variation was observed in spectra of ectocervical cells of women not taking any form of oral contraception. Ectocervical cells from women taking monophasic contraception and endocervical cells from both groupsdid not display the same degree of variation. Principal component analysis revealed that, although there is cyclical variation,these cells are normal and discrimination between histologically normal and abnormal (high-grade dysplasia) cells wasmaintained. # 2002 Elsevier Science B.V. All rights reserved. Keywords:  Infrared microspectroscopy; Ectocervical cells; Cyclical variation 1. Introduction Infrared spectroscopy has been extensively appliedto study changes at the molecular level of varioushuman cancers. Several groups have investigated theuse of IR spectroscopy in the diagnosis of colon [1–3],cervical [4–14], lung [15,16], and liver [17] cancers.The initial work in the application of infrared spec-troscopy for the detection of cervical cancer wasundertaken by Wong et al. [4,5]. Several observabledifferenceswerefoundbetweentheexfoliatedcervicalcells from women with normal cytology compared tothose with dysplastic cytology. The most notabledifferences were changes in the intensity of bands at1025, 1047, 1082, 1155, 1244 and 1303 cm  1 . Theratio of the peak intensities of bands at 1025 cm  1 (glycogen)and1082 cm  1 ( n s PO 2  ofphosphodiestergroups of nucleic acids) were found to differ greatlybetween normal and malignant cells. These findingswerefurthersubstantiatedandfoundtobeapplicabletomalignant cervical tissue [18].Fung et al. [19] compared FT-IR spectroscopy inthe screening of cervical cells with conventionalPapanicolaou smears using colposcopically directedbiopsy as the ‘‘gold standard’’. Specificity and sensi-tivity were reported for FT-IR (98.8 and 98.6%) andfor the Pap smear (90.5 and 86.8%). Infrared spectrawere classified as abnormal if they contained any of thespectralfeaturesidentifiedbyWongandco-workers[5,8].Although these findings indicate that infrared spec-troscopy is a powerful tool to discriminate betweennormal and malignant cervical cells, it is becomingincreasingly apparent that there may be other factorscontributing to the spectral changes assumed to bearising from neoplastic processes and malignancy.Possible contributing factors and/or confounding vari- Vibrational Spectroscopy 28 (2002) 167–175 * Corresponding author. Tel.:  þ 61-3990-54525;fax:  þ 61-3990-54597. E-mail address : d.mcnaughton@sci.monash.edu.au(D. McNaughton).0924-2031/02/$ – see front matter # 2002 Elsevier Science B.V. All rights reserved.PII: S0924-2031(01)00155-2  ables have been identi fi ed and include benign cellularchanges (BCC), speci fi cally metaplasia and in fl am-mation [8], erythrocytes [20], lymphocytes [11,20],endocervical cells [20] and mucins [11,12,20].Recently Diem ’ s group [13] conducted a series of experiments demonstrating that infrared spectroscopycould be used to monitor maturation and differentia-tion in cervical squamous epithelium. The observedspectral differences between the basal, parabasal,intermediate and super fi cial layers of the squamousepitheliumarosemainlyinthe1200 – 900 cm  1 region.The spectral differences observed showed an increasein glycogen concentration towards the surface, i.e. ascells matured from the basal layer they accumulatedmore glycogen. Differences were also noted in theamide I/amide II ratio, believed to be a result of nucleic acid contributions. Despite these differences,Cohenford and Rigas [10] found that the spectra of cytologically normal intermediate and super fi cialsquamous cells from women with dysplasia or cancerwere different from the cells of normal women.Multivariate statistics have been utilised by severalgroups to obtain a separation between the infraredspectra of normal, dysplastic and malignant samples.Wood et al. [7] used principal components analysis(PCA) to achieve a separation between the infra-red spectra of normal and dysplastic cells, whileCohenford et al. [9] employed principal componentregression (PCR) to achieve a separation betweennormal and malignant cervical cells. Romeo et al.[21] used PCA coupled with arti fi cial neural networks(ANNs) to classify unknown dysplastic and normalsamples.Morphologically there are many changes occurringin cervical epithelium as a direct result of hormonalstimulation from the menstrual and ovarian cycles.Cervical squamous epithelial cells accumulate glyco-gen as a process of maturation, the concentration of which is hormone dependent, peaking around ovula-tion. Giventhe  fi ndings of Diem ’ s group [13], outlinedabove, it would seem likely that the infrared spectra of cervical cells sampled throughout the menstrual cyclewould exhibit spectral differences.In this study we apply IR spectroscopy to monitorthe changes occurring during the hormonal cycle. Weinvestigated the hormonal in fl uences of cervicalepithelium throughout the menstrual cycle, obtainingspectrafromvolunteersonaweeklybasistodetermineif the cytological changes brought about by hormonalin fl uences could be manifested spectroscopically, andalso to determine if any changes could possibly con-found diagnosis. 2. Materials and methods 2.1. Participants Participants in this study were required to bepre-menopausal non-smokers with a history of normalPap smears, the most recent within the last 12 months.Because nicotine has been found to affect cellularproliferation of the cervix [22], smokers were notincluded in this study to reduce the number of con-tributing variables. Inclusion in the study was limitedto women taking no form of oral contraception andwomen taking monophasic contraception. Women nottaking oral contraception were required to have fourcervical smears, each cycle corresponding to post-menstrual,preovulatory,postovulatoryandpremenstr-ual phases of the menstrual cycle. These women werealso required to have blood taken once each cycle(postovulatory). Progesterone assays were undertakento ensure that ovulation had occurred and that thewomen had functional menstrual cycles. Cervicalsmears of women taking monophasic contraceptionwere taken on a weekly basis. 2.2. Sample collection Cervical cells were taken from the transformationzone of the cervix with both an Ayre spatula to ensurecollection of ectocervical cells, and a Cytobrush TM (MEDSCAND, Hollywood, FL, USA) to ensure col-lection of endocervical cells. Sampling instrumentswere agitated in separate 50 cm 3 centrifuge tubescontaining 10 cm 3 absolute ethanol to collect the cellsand then stored at   70  8 C until required. 2.3. Sample preparation Samples were centrifuged at 2500 rpm for 10 minand the ethanol supernatant removed with an auto-mated pipette leaving a cellular pellet. Ultra-cleanwater was added and the tubes were then vortexedto re-suspend and clean the cellular material. This 168  M.J. Romeo et al./Vibrational Spectroscopy 28 (2002) 167–175  washing procedure was carried out three times and thecellular material was then pipetted into a KRS-5multicavity cell [7] and desiccated under vacuum. 2.4. Infrared microspectroscopy Following removal of the aluminium plates used toform sample wells, the KRS-5 multicavity IR cell [7]containing 14 cervical samples was placed on thesampling platform of a Perkin-Elmer IR microscopecoupled to a Perkin-Elmer 1600 spectrophotometer. Aminimum of six transmission spectra were recordedfor each sample with the knife-edge aperture reducedto 50 m m  50  m m. For each spectrum, 16 scans wereco-added at a resolution of 8 cm  1 , with a totalrecording time for each spectrum of 20 s. 2.5. Data treatment  Infrared spectra in JCAMP format were transferredvia a Macintosh computer onto an OPUS (BrukerMesstechnik, Karlsruhe, Germany) operating plat-form. The spectra from each sample were re-scaled,baseline corrected, normalised to the amide I peak (1650 cm  1 ), averaged and then converted into anASCII format recognised by  Unscrambler II   (CAMOASA, Oslo, Norway).  Unscrambler   enabled visualinspection of the spectra and multivariate statisticalanalysis. Spectra that had absorbance greater thanunity (prior to normalisation) were discarded becauseof nonlinearity effects. Spectra with low signal-to-noise were also discarded. 2.6. Blood preparation Blood samples (10 cm 3 ) were centrifuged at2500 rpm for 15 min. Centrifugation separated theblood into three components: red blood cells, whiteblood cells, and serum. The serum was pipetted into a10 cm 3 centrifuge tube and frozen until the assay wasperformed. 3. Results and discussion Eleven non-smoking women participated in thisstudy for periods of between 4 and 12 weeks.Table 1 summarises the length of participation. 3.1. Progesterone assay All of the serum samples collected from partici-pants not on any form of contraception showed pro-gesterone, indicating that ovulation had occurred andthe women had functional ovulatory cycles. 3.2. Contamination A high proportion of the ectocervical cell sampledeposits were covered by a thin, white substance. Theinfrared spectra of these deposits, shown in Fig. 1,appeared unusual and were characterised by a doubletat 1053 and 1036 cm  1 , and peaks at 1108, 1160,1235, 1323 and 1730 cm  1 .The srcin of this contaminant was thought to bearising from loose  fi bres on the Ayre spatula, removedby agitation of the instrument in ethanol to collectcellular material. An Ayre spatula was vigorouslyagitated in a solution of absolute ethanol and thissolution was centrifuged, the resulting pellet pipettedinto a KRS-5 infrared cell and desiccated undervacuum. The resulting spectrum, also shown inFig. 1, showed similarities with the contaminatedectocervical cell spectrum, with peaks at 1033,1108, 1157, 1232, 1323 and 1730 cm  1 .Sampling instruments were initially kept in thecentrifuge tubes to maximise cell collection, however,after the contaminant was identi fi ed as Ayre spatuladebris, sampling instruments were only brie fl y agi-tated in the ethanol solution immediately after collec-tion and then discarded.A second source of contamination was mixed popu-lationsofectocervicalandendocervicalcells.Cervicalsmears are obtained from the transformation zone of the cervix, which is the area of the cervix where the Table 1Summary of participation in the studyNumber of womenLength of participation (weeks)Oralcontraception2 12 Monophasic2 12 None1 8 Monophasic1 8 None3 4 Monophasic2 4 None  M.J. Romeo et al./Vibrational Spectroscopy 28 (2002) 167   –  175  169  squamous epithelium (ectocervical) and columnarepithelium (endocervical) meet and where neoplasiais likely to occur [23]. It is sometimes dif  fi cult tosample squamous and columnar cells separatelybecause the transformation zone may not always bevisible to the person taking the smear. The location of the transformation zone changes throughout the life-time of a female and depends on age, reproductivestatus and pregnancy [24]. Consequently, sampleswhich showed infrared spectra characteristic of bothectocervical and endocervical epithelium were dis-carded to minimise the chance of spectra resultingfrom a mixture of these two components confoundingthe affects occurring from cyclical changes. 3.3. Spectral subtraction The contribution of Ayre spatula contamination insome of the ectocervical cell spectra was removedby spectral subtraction using Grams 32 software(Warsash, Sydney, Australia). A direct subtractionof the spectrum obtained directly from an Ayre spatulaagitation was not successful due to two reasons.Firstly, the absolute intensity of the spatula spectrumcompared to the ectocervical cell spectra is very smalland subtraction over a large wavenumber region wasnot possible with a single scaling factor. Secondly, thecontaminant within the smear sample appears to haveslight differences to the spectrum of the spatula debrisalone. Consistent spectral subtraction was achievedby subtraction of a non-contaminated ectocervical cellspectrum from a contaminated spectrum using theelimination of the 1730 cm  1 peak to determine thesubtractionfactor.Whilethismethod islessthanideal,the same spectrum was used for all the subtractions. 3.4. Ectocervical cells Ectocervicalcellsarecharacterisedbythefollowing:amide I and amide II bands at 1651 and 1544 cm  1 ,respectively:peaksat1453and1378 cm  1 arisingfromthe asymmetric and symmetric deformation mode of the methyl groups in proteins, a very weak amide IIIband at 1318 cm  1 , asymmetric and symmetric phos-phate stretches at 1242 and 1081 cm  1 , respectively, aband at 1154 cm  1 arising from C – O stretching vibra-tion of proteins and carbohydrates, and a peak at1025 cm  1 arising from the  – CH 2 OH stretching vibra-tion of glycogen.The infrared spectra of ectocervical cells exhibiteda consistent variation throughout the cycle and anexample is shown in Fig. 2. The main differencesare observed in the carbohydrate region (1200 – 1000 cm  1 ). Increases in the intensities of the peaksattributable to glycogen would be expected towardsthe mid-cycle days of the cycle, as a result of glycogenaccumulation in the intermediate and super fi cial cells[25]. This increase is clearly seen with the infraredspectra resulting from days 8 and 12 showing markedincreases in the 1025 cm  1 band, attributed to glyco-gen, compared to days 19 and 26. Glycogen concen-trations are expected to peak around ovulation, whichin this 30-day cycle would occur at day 16.Of the 17 cycles of women not taking oral contra-ception examined in this study, all but two cyclesexhibited this spectral pattern, i.e. the glycogen bandat 1025 cm  1 increased towards the mid-cycle daysand decreased following ovulation.The consistency of these changes is also observedwhen similar days from different cycles in the samesubject are examined. Fig. 3 shows the mid-cycle days Fig. 1. (a) Infrared spectra of contaminated ectocervical cell (A),uncontaminated ectocervical cell (B) and Ayre spatula (C). Spectraare baseline corrected but not normalised; (b) shows an enlargedspectrum of the Ayre spatula.170  M.J. Romeo et al./Vibrational Spectroscopy 28 (2002) 167   –  175  from a woman not taking oral contraception. It isknown that cycle length both between and withinwomen can differ substantially [26] and the slightdifferences in these spectra are possibly due to dif-ferent cycle lengths or simply to normal biologicalvariation.Fig. 4 shows the infrared spectra of ectocervicalcells collected from a woman taking monophasic oralcontraception. As expected the spectra do not exhibitlarge variation throughout the cycle, with small dif-ferences observed in the carbohydrate region. All of the cycles recorded from the monophasic participantsshowed similar spectral patterns, although at the endof the cycle glycogen levels were generally found todecrease. This may be due to the withdrawal of monophasic contraception to allow for menstruation.While it is obvious that there is a large degree of variability in cervical cells throughout the menstrualcycle, these cells are essentially normal and would beexpected to group with normal cervical cells in a PCAscores plot.PCA was performed on 240 infrared spectra withknown biopsy results from one of our other studies.Fig. 5A shows a clear separation between normal anddysplasticsamplesonaPC1versusPC2scoresplotforthis data alone. Sixty six infrared spectra from allparticipantsinthisstudywereaddedtothedatasetandPCAwas performed a second time. The resulting PC1 Fig. 2. Infrared spectra obtained from a woman, not taking oral contraception, over one cycle. The length of this cycle was 30 days.Fig. 3. Infrared spectra of a woman, not taking oral contraception, collected over four cycles.  M.J. Romeo et al./Vibrational Spectroscopy 28 (2002) 167   –  175  171
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks