144 / 236
Hauser et al.
FIGURE 1.
Phyla and genera of ethmoid bacterial communities over time. Bars represent the average relative abundances of the phyla and genera present
in the ethmoid samples at 0, 2, and 6 weeks postoperatively. rRNA
=
ribosomal RNA.
Results
Thirteen subjects met the criteria for inclusion in the study
and 12 subjects had adequate sampling for analysis of each
subsite and time point. The mean age of our subjects was
56.6 years (range, 28–77 years). Sixty-one percent (61%)
had polyps and asthma and 46% were undergoing revision
surgery. Subjects had moderate disease severity based aver-
age Lund-Mackay computed tomography (CT) and Lund-
Kennedy endoscopy scores of 10.7
±
4.6 and 5.7
±
2.2,
respectively. The average relative abundances of the phyla
and genera present in the ethmoid samples at 0, 2, and 6
weeks postoperatively is depicted in Figure 1.
Overall bacterial burden was estimated using total qPCR
16S bacterial gene copy numbers. The mean log
10
of the
gene copy number of the ethmoid samples at the time of
surgery, 2 weeks, and 6 weeks postoperatively were 3.53,
4.30, and 3.47, respectively (Fig. 2). The bacterial bur-
den of the ethmoid swab was significantly higher at the
2-week time point (immediately after completion of antibi-
otics) than at 6 weeks postoperatively (
p
=
0.03, by 2-tailed
paired
t
test).
The Morisita-Horn beta-diversity index (M-H) was used
to compare the similarities of microbiomes between pairs
of samples (a value of 0 indicates no similarity, whereas
a value of 1 indicates identical microbial community com-
position). All samples (anterior nares, ethmoid, and na-
sopharynx at the time of surgery, and ethmoid 2 weeks
postoperatively) were compared to the 6-week postopera-
tive population (Fig. 3). Additionally, the 2-week ethmoid
postoperative sample was compared to the ethmoid sam-
ple at the time of surgery. The 6-week postoperative sample
was most closely matched to the anterior nares and ethmoid
samples taken at the time of surgery (mean M-H
=
0.58 and
0.59, respectively), suggesting that these sites may have the
most impact on bacterial repopulation postoperatively. The
nasopharynx was the least similar to the bacterial makeup
of the ethmoid specimens 6 weeks postoperatively (M-H
=
0.29 and 0.18, respectively), suggesting that it likely to
does not contribute to postoperative bacterial recoloniza-
tion. The 6-week ethmoid samples were less similar to the
2-week samples (M-H
=
0.40) than to the baseline ethmoid
samples (M-H
=
0.59).
Principal coordinates analysis (PCoA) plots illustrate that
the ethmoid microbiota shifted after surgery and antibiotic
administration but returned toward the original baseline
in many of the patients (Fig. 4). The results of a PER-
MANOVA test indicated that variability in microbiomes
was driven more by intersubject variation (
p
=
0.0004)
than by time-point of sampling (
p
=
0.5) when both vari-
ables were entered into the regression as predictor variables.
None of the clinical variables that we assessed (allergies,
asthma, polyposis, purulence) were significant predictors of
microbiome structure after adjusting for subject and sam-
pling time-point.
Among the ethmoid samples, no significant differences in
OTU richness (ie, the number of OTUs per subject) were
noted between baseline, 2-week, or 6-week samples. How-
ever, the complexity of 2-week ethmoid samples trended
toward being significantly lower than the 6-week samples
(mean Shannon diversity index of 1.9
±
1.2 vs 2.9
±
1.1, re-
spectively;
p
=
0.09); baseline samples were of an interme-
diate complexity (2.4
±
0.8), but did not differ significantly
from either the 2-week or 6-week samples.
To evaluate for potential contamination during sampling,
average Morisita-Horn similarities were calculated com-
paring the anterior nares, the ethmoid, and the nasophar-
ynx samples at the time of surgery. The anterior nares and
ethmoid samples were the most similar (M-H 0.81); how-
ever, this is not surprising given these were both subse-
quently found to be potential sources for repopulation of
the ethmoid cavity. The bacterial population of the na-
sopharynx was not similar to either the ethmoid cavity or
the anterior nares (M-H 0.46, 0.43), indicating there was
likely little contamination of the nasopharynx swab when
passing through the proximal sinonasal cavity.
Discussion
Recent work has begun to elucidate the composition and
biodiversity of the sinus microbiome in both health and
International Forum of Allergy & Rhinology, Vol. 6, No. 1, January 2016
122