Rapid identification of bacteria in blood cultures by mass ...

Downloaded from jcp.bmj.com on July 16, 2014 - Published by group.bmj.com

PostScript

Figure 4 Histological examination showed a fairly circumscribed lobulated lesion with focal infiltration through dense fibrotendinous tissue (A; H&E, ×2). The tumour cells were composed of highly cellular and cytologically atypical spindled (B; H&E, ×20) and epithelioid cells (C; H&E, ×10) with focal perivascular growth pattern (D, H&E, ×4). Focally there were aggregates of multivacuolated lipoblasts with hyperchromatic scalloped nuclei (E; H&E ×20). S100 immunohistochemical staining was negative (F; S100, ×20). composed entirely of sheets of pleomorphic, multivacuolated lipoblasts. High mitotic activity and extensive necrosis is usually present. Diagnosis of PLS is mainly dependant on the morphology. Immunohistochemical staining is mainly used to exclude the differential diagnosis of PLS, such as INI-1 loss in epithelioid sarcoma, positive melanocytic markers in metastatic melanoma and clear cell sarcoma and positive muscle markers in leiomyosarcoma and rhabdomyosarcoma. S100 staining is patchily positive in some and negative in other studies,4 5 as seen in this case report. The differentiation between DLS, pleomorphic MLS and PLS may be difficult. However with the advent of molecular cytogenetic analysis, the lack of MDM2 homolog fluorescent in situ hybridisation amplification and FUSDDIT3 or EWSR1-DDIT3 fusion aids in excluding the diagnosis of DLS and pleomorphic MLS. respectively.7 8 PLS is associated with poor survival, with a local recurrence rate, metastatic rate and overall mortality range between 40% and 50%. However, studies have showed that limb-based PLS with local radiotherapy and wide excision or amputation significantly reduce the rate of local recurrence. Histological features such as tumour size ≤5 cm, lower mitotic rate (≤10 per 10 high power field (HPF)) and lack of necrosis are associated with significantly improved metastasis-free survival.4 6 In summary, PLS is a rare high grade sarcoma arising mainly on the limb girdles, which may appear well circumscribed and benign on gross appearance, but show a wide range of cytologically atypical appearance on histological examination. Extensive sampling is recommended for identification J Clin Pathol August 2014 Vol 67 No 8

of lipoblasts which is a prerequisite for diagnosis. Immunohistochemical staining and molecular cytogenetic analysis is mainly useful for the exclusion of differential diagnosis. We report here the first known case in the English literature of PLS arising from the hand. Further follow-up of this case and other subsequent similar cases is recommended to determine if the prognosis is similar to that of PLS arising elsewhere.

J Clin Pathol 2014;67:741–743. doi:10.1136/jclinpath-2014-202390

REFERENCES 1 2

3

4

Po Yin Tang, Rafay Azhar, Sittampalam Kesavan Department of Pathology, Singapore General Hospital, Singapore 5

Correspondence to Dr Sittampalam Kesavan, Department of Pathology, Singapore General Hospital, 20 College Road, Level 10, Diagnostics Tower, The Academia, Singapore 169856, Singapore; [email protected]

6

Correction notice Figure 4 has been corrected since published Online First.

7

Acknowledgements The authors thank Prof BH Tan for permission to report this case and Prof Christopher DM Fletcher for reviewing this case. Contributors PYT collected patient data, and drafted and revised the paper. She is one of the guarantors. RA diagnosed the case and revised the draft paper. SK aided in the diagnosis of the case, initiated the writing of the report and revised the draft paper. He is the other guarantor. Competing interests None. Provenance and peer review Not commissioned; internally peer reviewed.

8

McPhee M, McGrath BE, Zhang P, et al. Soft tissue sarcoma of the hand. J Hand Surg 1999;24:1001–7. Fletcher DM, Bridge J, Hogendoorn PCW, et al. WHO Classification of tumours of soft tissue and bone. 4th edn. World Health Organization, 2013. Azumi N, Curtis J, Kempson RL, et al. Atypical and malignant neoplasms showing lipomatous differentiation: a study of 111 cases. Am J Surg Pathol 1987;11:161–83. Gebhard S, Coindre JM, Michels JJ, et al. Pleomorphic liposarcoma: clinicopathologic, immunohistochemical, and follow-up analysis of 63 cases. A study from the French Federation of Cancer Centers Sarcoma Group. Am J Surg Pathol 2002;26:601–16. Wang L, Ren W, Zhou X, et al. Pleomorphic liposarcoma: a clinicopathological, immunohistochemical and molecular cytogenetic study of 32 additional cases. Pathol Int 2003;63:523–31. Ghadimi MP, Liu P, Peng T, et al. Pleomorphic liposarcoma: clinical observations and molecular variables. Cancer 2011;117:5359–69. Shimada S, Ishizawa T, Ishizawa K, et al. The value of MDM2 and CDK4 amplification levels using real-time polymerase chain reaction for the differential diagnosis of liposarcomas and their histologic mimickers. Hum Pathol 2006;37:1123–9. Alaggio R, Coffin CM, Weiss SW, et al. Liposarcomas in young patients: a study of 82 cases occurring in patients younger than 22 years of age. Am J Surg Path 2009;33:645–58.

Rapid identification of bacteria in blood cultures by mass-spectrometric analysis of volatiles INTRODUCTION

To cite Tang PY, Azhar R, Kesavan S. J Clin Pathol 2014;67:741–743. Received 28 April 2014 Revised 1 May 2014 Accepted 2 May 2014 Published Online First 28 May 2014

Blood cultures are routine tests to determine whether micro-organisms have entered the patient’s bloodstream. Automated systems, based on the detection of CO2 increase in the culture media, have considerably improved the screening efficiency for the 743


PostScript Table 1 Volatile compounds detected in real-time in the head-space of the three bacteria investigated

m/z 57.07028 59.04946 69.06999 71.08584 73.04711 73.06511 89.04203 89.05991 91.05756 95.0609 95.08561 97.06487 105.0547 105.073 105.0911 109.076 109.1012 113.0961 115.0754 119.0887 119.1066 127.1118 131.0887 142.0631 145.1043 145.1223 162.1489 163.0969 177.1485 203.1641 219.1745 233.2112 308.2796

Molecular formula

C3H7O C5H9 C5H11 C4H9O C4H9O2 C4H11S

C6H9O C4H9O3 C5H13S C5H13O2 C6H9N2 C7H13O C6H11O2 C6H15S C8H15O C7H15S C8H17S C8H17O2 C8H20NO2 C7H15O4 C9H21O3 C11H23O3 C13H29O3 C14H36O3N4

Staphylococcus aureus (n=15) mean (SD)

Escherichia coli (n=15) mean (SD)

Streptococcus pneumonia (n=12) mean (SD)

Medium (n=15) mean (SD)

p-Value (ANOVA)

1.46 (0.31) 6.03 (5.48) 5.53 (1.81) 0.94 (0.2) 29.08 (15.25) 16.05 (6.4) 1.9 (1.64) 16.37 (7.53) 57.46 (34.51) 0.37 (0.33) 2.28 (0.96) 6.47 (1.39) 4.13 (2.91) 61.17 (33.58) 0.89 (0.47) 2.9 (1.66) 4.41 (2.88) 8.78 (4.45) 75.58 (23.7) 46.51 (21.06) 5.35 (2.25) 10.14 (3.71) 13.49 (8.18) 1.73 (2.29) 3.84 (2.79) 9.92 (8.96) 3.33 (2.34) 25.35 (8.43) 1 (1.35) 0.99 (1.89) 0.51 (0.39) 2.15 (3.2) 6.32 (12.01)

1.45 (0.47) 6.57 (5.57) 7.47 (2.62) 2.32 (1.32) 20.82 (15.1) 30.47 (13.77) 0.87 (0.99) 29.43 (23.98) 40.69 (27.27) 0.64 (0.32) 2.19 (1.03) 6.69 (1.64) 9.17 (5.23) 28.99 (25.86) 1.42 (0.76) 3.86 (1.69) 10.47 (11.34) 16.48 (8.86) 77.72 (18.78) 69.7 (32.42) 5.31 (1.72) 19.12 (14.96) 11.5 (7.63) 5.61 (5.31) 3.6 (2.44) 28.41 (16.95) 5.87 (4.13) 17.83 (8.83) 3.18 (2.29) 2.3 (2.74) 1.41 (1.06) 7.36 (6.24) 36.31 (39.34)

1.89 (1.09) 3.41 (2.42) 8.82 (2.93) 1.19 (0.99) 23.55 (12.98) 30 (13.61) 2.01 (1.45) 22.06 (6.88) 32.04 (18.56) 0.42 (0.26) 2.68 (0.76) 8.37 (3.42) 10.85 (3.01) 63.24 (34.16) 1.22 (0.69) 3.74 (1.24) 11.17 (7.03) 18.26 (5.1) 90.1 (13.65) 2.42 (2.92) 6.3 (1.99) 20.14 (7.1) 19.66 (20.32) 5.35 (4.24) 3.14 (2.5) 28.85 (10.78) 5.46 (3.86) 13.32 (5.03) 3.74 (2.9) 2.42 (3.32) 1.76 (1.75) 8.45 (7.07) 30.44 (41.73)

1.11 (0.63) 3.81 (2.55) 5.48 (2.02) 0.51 (0.25) 32.27 (10.25) 13.79 (5.49) 2.98 (2.61) 12.34 (18.6) 67.07 (37.54) 0.25 (0.33) 1.85 (0.74) 4.12 (1.46) 0.43 (0.7) 73.26 (36.1) 0.5 (0.33) 0.95 (1.16) 2.53 (1.3) 3.99 (1.3) 37.04 (20.75) 2.4 (1.52) 2.57 (1.26) 6.29 (2.21) 16.57 (10.68) 0.1 (0.39) 5.51 (3.78) 3.42 (2.98) 2.92 (2.96) 29.7 (12.1) 0.18 (0.29) 0.71 (1.44) 0.31 (0.24) 0.18 (0.36) 5.78 (12.84)

1.9E-01 2.2E-01 4.9E-03 7.7E-04 3.0E-01 2.2E-03 6.2E-02 8.1E-02 6.7E-02 5.1E-02 3.6E-01 7.4E-02 1.5E-04 8.2E-03 9.3E-02 2.1E-01 5.5E-02 9.2E-04 1.4E-01 2.4E-08 3.7E-01 2.0E-02 2.5E-01 2.6E-02 7.8E-01 2.7E-04 1.2E-01 9.7E-04 5.3E-03 2.9E-01 1.8E-02 1.1E-02 4.4E-02

Average signal of compounds in italics was found to be higher than for the three pathogens investigated.

detection of bacteria.1 However, further identification of bacteria still requires timeconsuming culturing procedures. It has been suggested that along with CO2, bacterial cultures emit characteristic volatile organic compounds that may be valuable for characterisation.2 Recently, a number of technological developments have allowed the detection of trace gases with minimal or no sample preconcentration steps.3–5 In the current study, we investigated whether the volatiles emitted from bacterial cultures (mimicking routine clinical blood cultures) could be distinguished from each other in a rapid fashion by secondary electrospray ionisation-mass spectrometry (SESI-MS).6 7

METHODS 103 colony forming units were inoculated in 42 aerobic bottles: 15 with 744

Staphylococcus aureus ATCC 29213 strain, 15 with Escherichia coli ATCC 25922 and 12 Streptococcus pneumoniae ATCC 49619. They were incubated in the automatic instrument BacT/ALERT 3D (Biomerieux Clinical Diagnostic, France). The system automatically detects the positivity of the sample when the amount of CO2 produced by bacterial growth reaches a given threshold. The bacterial identity was confirmed by culturing according to UK guidelines. The headspace gas of the positive bottles was sampled with a 5 mL syringe. The gas was subsequently injected into a modified high-resolution mass spectrometer (Orbitrap, Thermo) located in the hospital premises. As a result, an immediate mass spectral fingerprint was obtained. The mass spectra were subjected to one-way analysis of variance (ANOVA) test

and principal component analysis (PCA). A k-nearest-neighbour (k-nn, k=1) classifier was employed to predict sample class in a leave-one-out cross-validation (LOOCV).

RESULTS An immediate mass spectral fingerprint was obtained from all samples. Thirty-three signals in the range 57–308 m/z were found to arise above the background upon injection of the gas. Table 1 lists their m/z values, molecular formula, mean (SD) and p value resulting from the ANOVA test. The average intensity of 27 out of the 33 compounds detected was higher in at least one of the bacterial strains than in the medium. Figure 1 shows an example of the real-time measurement of nine samples (three biological replicates of each bacteria strain) for six representative compounds. Note that each measurement requires a few seconds. Thus, in approximately 10 min, rich volatile fingerprints were obtained for the nine samples. Note how the signal rises upon the injection of the sample with a characteristic intensity, depending on the bacterial type, with a reasonable reproducibility for the three different biological replicates. For example, the compound at m/z 119.0887 (C6H15S) was present in S. aureus and E. coli, but not in S. pneumonia (figure 1A) and the medium (table 1). The corresponding PCA score plot (figure 2) of the mass spectral fingerprints reveals a clear clustering according to bacterial specimen.8 The LOOCV resulted in 40 out of 42 (95.2%) samples correctly classified. One S. aureus and one S. pneumoniae sample were misclassified as E. coli.

CONCLUSIONS This exploratory study suggests that it is possible to identify bacteria by analysing the gases produced in the headspace of blood cultures by mass spectrometry in a rapid fashion with no sample preparation. Consistently with recent real-time mass spectrometric studies of bacterial cultures,3 4 relatively low molecular weight volatiles like acetone (m/z 59), acetoin (m/z 89) and butanal (m/z 73) were found in the cultures’ headspace. In addition to this, SESI-MS captures rich fingerprints of volatiles expanding well beyond 300 Da.9 This data suggests that such high molecular weight vapours may be key to deliver high accuracy in pattern recognition of bacteria. For example, the ion detected at m/z 308.2796 (C14H36O3N4), was found to be the second largest relative contributor in PC3, which essentially separates E. coli from S. aureus and S. pneumonia (see loadings plot in online supplementary figure S1). However, this pilot study has J Clin Pathol August 2014 Vol 67 No 8


PostScript 4

Pulmonary Division, University Hospital Zurich, Zurich, Switzerland 5 Experimental Medicine Department, School of Medicine, University of Milano-Bicocca, Monza, Italy 6 National Research Council-Institute for Biomedical Technologies, Segrate, Italy * Current address: Department of Chemistry and Applied Biosciences, ETH Zurich, Zurich, Switzerland Correspondence to Dr Pablo Martinez-Lozano Sinues, ETH Zurich, Department of Chemistry and Applied Biosciences, Vladimir-Prelog-Weg 3, Zurich CH-8093, Switzerland; [email protected] Contributors SC, PB and PM-LS designed the study. CB, SC, MRS and PM-LS conducted the experiments. PM-LS analysed the data. SC, GP, MK, MRS, PB and PM-LS wrote the manuscript. Funding This research was supported by a Marie Curie Grant (PMLS) within the 7th European Community Framework Programme (PIEF-GA-2008220511). Competing interests None. Provenance and peer review Not commissioned; externally peer reviewed. ▸ Additional material is published online only. To view please visit the journal online (http://dx.doi.org/ 10.1136/jclinpath-2014-202301).

Figure 1 The headspace of commercial blood cultures can be sampled and mass analysed in a few seconds, providing an indication of the pathogen present. The plot displays the intensity as a function of time of six exemplary compounds detected in the headspace of the bacterial cultures. Three biological replicates of Staphylococcus aureus (Sa), Escherichia Coli (Ec) and Streptococcus pneumonia (Sp) were injected at minutes 1.1, 2.2, 3.3; 5.4, 6.1, 6.7 and 9.1, 10.5, 11.3, respectively (A) m/z 119.0887 (C6H15S); (B) m/z 163.0969 (C7H15O4); (C) m/z 73.06511 (C4H9O); (D) m/z 105.073 (C5H13S); (E) m/z 219.1745; (F) m/z 308.2796 (C14H36O3N4). some limitations. For example, during this exploratory phase, we wanted to determine the feasibility of this approach by using pure bacterial standards with relatively high colony-forming units. Further research is needed to confirm these findings in expectedly more complex matrices like blood cultures from patients.

1

2

Claudia Ballabio, Simone Cristoni, Giovanni Puccio,3 Malcolm Kohler,4 Maria Roberta Sala,1 Paolo Brambilla,1,5 Pablo Martinez-Lozano Sinues6,* 1

Clinical Pathology Department, Desio Hospital, Desio, Italy 2 ISB Ion Source & Biotechnologies, Milan, Italy 3 Emmanuele Scientific Research Association, Palermo, Italy

To cite Ballabio C, Cristoni S, Puccio G, et al. J Clin Pathol 2014;67:743–746. Received 17 March 2014 Revised 11 April 2014 Accepted 16 April 2014 Published Online First 9 May 2014 J Clin Pathol 2014;67:743–746. doi:10.1136/jclinpath-2014-202301

REFERENCES 1

2

3

4

5

6

Figure 2 Proof-of-principle of rapid characterisation of three bacterial strains grown in blood bacterial culture vials. Score plot of principal component analysis of the mass spectral fingerprint of the volatiles emitted by three different bacterial agents: Staphylococcus aureus (n=15 samples), Escherichia coli (n=15) and Streptococcus pneumonia (n=12). Clustering according to pathogen is evident, suggesting that before disposal of the positive vials, the analysis of its headspace may be a valuable source of information. Note that each of the measurements shown in the graph requires just a few seconds. J Clin Pathol August 2014 Vol 67 No 8

7

Thorpe TC, Wilson ML, Turner JE, et al. BacT/Alert: an automated colorimetric microbial detection system. J Clin Microbiol 1990;28:1608–12. Bos LDJ, Sterk PJ, Schultz MJ. Volatile metabolites of pathogens: a systematic review. PLoS Pathog 2013;9: e1003311. Bunge M, Araghipour N, Mikoviny T, et al. On-line monitoring of microbial volatile metabolites by proton transfer reaction-mass spectrometry. Appl Environ Microbiol 2008;74:2179–86. Sovova K, Cepl J, Markos A, et al. Real time monitoring of population dynamics in concurrent bacterial growth using SIFT-MS quantification of volatile metabolites. Analyst 2013;138: 4795–801. Chippendale TWE, Gilchrist FJ, Spanel P, et al. Quantification by SIFT-MS of volatile compounds emitted by in vitro cultures of S aureus, S pneumoniae and H influenzae isolated from patients with respiratory diseases. Anal Methods 2014;6:2460–72. Dillon LA, Stone VN, Croasdell LA, et al. Optimisation of secondary electrospray ionisation (SESI) for the trace determination of gas-phase volatile organic compounds. Analyst 2010;135:306–14. Martínez-Lozano P, Rus J, Fernández de la Mora G, et al. Secondary Electrospray Ionization (SESI) of ambient vapors for explosive detection at concentrations below parts per trillion. J Am Soc Mass Spectrom 2009;20:287–94.

745


PostScript 8

9

Zhu J, Bean HD, Kuo Y-M, et al. Fast detection of volatile organic compounds from bacterial cultures by secondary electrospray ionization-mass spectrometry. J Clin Microbiol 2010;48:4426–31. Martínez-Lozano P, Fernández de la Mora J. Electrospray ionization of volatiles in breath. Int J Mass spectrom 2007;265:68–72.

CORRESPONDENCE

Peritoneal dissemination of a gastric API2-MALT1-positive MALT lymphoma BACKGROUND The majority of mucosa-associated lymphoid tissue (MALT) lymphomas arises in the stomach but a variety of extranodal sites inside or outside the gastrointestinal (GI) tract can be afflicted.1 Gastric MALT lymphomas most frequently spread to a GI site, MALT lymphomas of extragastric origin rather disseminate to a non-GI site. Moreover, gastric MALT lymphomas disseminate less frequently (up to 25%) than extragastric MALT lymphomas (up to 46%).2 It has been reported that at diagnosis up to a third of all MALT lymphomas have disseminated with involvement of one or multiple MALT-containing organs or non-MALT-containing organs.1 The lymph nodes, bone marrow, spleen and liver are common non-MALT sites but also dissemination to the pleura has been described.3 However, involvement of the peritoneum has not been reported before.

CASE PRESENTATION A patient with an unremarkable social and family history was referred to us for therapy of a newly diagnosed rectal tumour suspicious for cancer. Staging confirmed a large, obstructing tumour of the middle rectum, which showed high-grade dysplasia and atypical glands suspicious for invasive cancer on histological examination (category 4.3 of the revised

Vienna classification). Although, signs of locoregional lymph node metastases were missing, we found multiple small mesenteric lesions and ascites in addition to a thickening of the entire stomach wall (figure 1A) and suspiciously enlarged retrocrural and upper abdominal lymph nodes on a CT scan. Subsequent oesophagogastroduodenoscopy revealed multiple peptic ulcers with mass lesions of the stomach. Histologically, gastric biopsies presented an infiltration of a MALT lymphoma with small to intermediate B cells with pale cytoplasm and destruction of gastric corpus glands with typical lymphoepithelial lesions (figure 2A). Immunohistochemically, lymphoid cells were positive for CD20 (figure 2B). A Helicobacter pylori infection could not be detected. Infiltration of another B cell lymphoma, especially a diffuse large B cell lymphoma was excluded (low proliferation-index of maximum 20%; negative reaction of CD10, Cyclin-D1 and CD23). Taken together, staging findings prompted us to assume that either a late stage (disseminated) MALT lymphoma or peritoneal carcinomatosis was present. To solve this diagnostic dilemma, we decided to perform a staging laparoscopy that showed a low amount of ascites and white lesions to the parietal peritoneum (figure 1B) of the lower abdominal wall. Ascites cytology was unremarkable. Histological examination of the peritoneal specimens excluded carcinomatosis, but exhibited peritoneal infiltration by CD20 positive B-lymphoid cells with the same cytomorphology as the lymphoma in the stomach (figure 2C, D). Subsequently, the patient underwent an anterior rectal resection with an uneventful postoperative course except for a small subcutaneous wound infection. The histopathological examination of the resected specimen revealed a stage Ia (Union

international contre le cancer) rectal adenocarcinoma that developed from a large tubulovillous adenoma with highgrade dysplasia. Moreover, the mesenteric adipose tissue and lymph nodes presented an infiltration of the prediagnosed lowgrade MALT lymphoma (figure 2E, F). Monoclonal immunoglobulin heavy chain gene rearrangements were evident in the adipose tissue and the stomach with the same monoclonal pattern in three conversed framework regions of the IgH gene and the joining ( J-) region (inVivoScribe Mix A IgH (310–360 bp), Mix B IgH (250–295 bp) and Mix C IgH (100– 170 bp) (figure 3D). Fluorescence in situ hybridisation revealed an aberrant clone with API2-MALT1-fusion indicating translocation t(11;18)(q21;q21) (figure 3A, B). Notably, gains of chromosomes 3 and 12 were detected using probes for centromere 3 and 3q27 (BCL6, figure 3C) as well as 12p13 (ETV6) and 12q13 (DDIT3). The patient was discharged 9 days after surgery. The follow-up visits confirmed good postoperative convalescence. Consequently, the patient was started on six cycles of chemotherapy with bendamustine plus rituximab.

DISCUSSION Patients with colorectal cancer have a higher incidence of double primary malignancies than the general population. In a retrospective analysis of 2,301 patients with colorectal cancer, Yun et al4 identified 145 patients (6.3%) with double primary malignancies, most of which were located in the stomach (40%). Unfortunately, the authors did not provide information regarding the presence of MALT lymphomas in their cohort. An advanced Medline search revealed single case reports of primary lymphoma of the stomach or colon and synchronous or metachronous adenocarcinomas.5 However, no reports of GI

Figure 1 (A) Abdominal CT scan depicts a diffuse wall thickening of the stomach and perihepatic ascites. (B) Laparoscopic image of the anterior abdominal wall showing lymphoma dissemination to the parietal peritoneum.

746

J Clin Pathol August 2014 Vol 67 No 8


Rapid identification of bacteria in blood cultures by mass-spectrometric analysis of volatiles Claudia Ballabio, Simone Cristoni, Giovanni Puccio, et al. J Clin Pathol 2014 67: 743-746 originally published online May 9, 2014

doi: 10.1136/jclinpath-2014-202301

Updated information and services can be found at: http://jcp.bmj.com/content/67/8/743.full.html

These include:

Data Supplement

"Supplementary Data" http://jcp.bmj.com/content/suppl/2014/05/09/jclinpath-2014-202301.DC1.html

References

This article cites 9 articles, 3 of which can be accessed free at: http://jcp.bmj.com/content/67/8/743.full.html#ref-list-1

Email alerting service

Receive free email alerts when new articles cite this article. Sign up in the box at the top right corner of the online article.

Notes

To request permissions go to: http://group.bmj.com/group/rights-licensing/permissions

To order reprints go to: http://journals.bmj.com/cgi/reprintform

To subscribe to BMJ go to: http://group.bmj.com/subscribe/

Figure S1. (a) Score plot of PC2 vs. PC3 suggests a clustering according to bacterial strain. The corresponding loadings plots of PC2 (b) and PC3 (c).