Critical appraisal of articles about lung microbiota in
early CF patients (1, 2)
Study design
The first article (1)
is a cohort study looking at infants diagnosed with CF after neonatal
screening. Bronchoalveolar lavages (BAL) were performed at the time of study entry, if hospitalized, and
annually after entry. These samples were then processed and analysed for bacterial genes
using 16-S sequencing as well as quantitative cultures and measurement of
different inflammatory markers, especifically IL-8 and neutrophil electalase. In
the end, a total of 95 samples from 48 subjects could be successfully sequenced,
and successive samples could be obtained in 27 of these subjects. Data from
medical records were obtained, including the best FEV1 at 6 years of
age, which was subsequently recalculated using Global Lung Initiative 2012
reference values to measure percentage of expected.(3) Information about whether they were symptomatic at the time of BAL was also obtained from records.
It is worth mentioning the time between the collection of
samples (1992 – 2001) and the date of publication (2017). The first study does
a secondary analysis on these samples that had been stored before for other
purposes. However, measurements have been taken
to ensure the samples are as representative as possible. Results were correlated to cultures made when the BAL was performed (so
between 1992 – 2001) and to the inflammation values that were measured. In the end the researchers could not find a relationship
between the microbiome and pulmonary function at the age of 6, but did find a link
between certain pathogens and inflammation.
The second article is a cross sectional study comparing the
lung microbiota of healthy children with those of currently stable children diagnosed with CF.(2)
Like the other study, it uses BAL fluid samples and 16S analysis for this as well as a
standard culture. Unlike the other study though, this study has a control group and
excluded patients if they had received antibiotics. It mainly concluded that there
were significant differences between the lung microbiome of children with and
without CF. It also suggest pseudomonas and haemophilus might be competitors. What in my opinion is very strong about the second study is that it
goes to great lengths to describe the methods applied in addition to various techniques that they used to prevent not only different errors in data collection, but also forms of bias.
Limitations of
study design
The first article
states that amplification failed in 53 samples, possibly due to low load or the
age of the samples. It is possible of course that this affected the outcome, because these samples contained important data which was not able to be measured due to insufficient sensitivity. One very important aspect to note is that to preserve the BAL samples,
they were frozen at -70C, this can significantly affect the results because
different bacterial species may have different resistances to being frozen.(4)
Now this should only affect cultures and not PCR because sequencing based methods are more likely to still produce a signal even if bacteria are dead. Nevertheless, mentioning this is important because delays or variations in freezing may give a less accurate representation of actual microbiota. (5) The first article does not describe how fast the samples have been stored.
The second article had a control
group, unlike the first article, which makes it possible to appreciate the differences between the lung
microbiota of healthy versus CF patients. Its sample size, however, was significantly smaller in this article with 13 patients and 9 controls making the impact far less
powerful. It is also a cross-sectional study as opposed to the other study which was longitudinal, which has the disadvantage that
nothing can be reliably said about change in the microbiome over time and day to day variability. This
article does not find a link between age and difference in microbiome diversity, which could be due to its cross-sectional design however. If you want to make a statement about a change in microbiome diversity with age, comparing the microbiome of younger children with CF to older children with CF is not going to be as reliable as following these children over time and measuring changes in microbiome diversity like a longitudinal study does.
16-S sequencing has been used in both studies and can be
expected to be able to find all the bacteria found in the lungs, given the
bacterial load is high enough. However it might not be specific enough to classify
bacteria past the genus level and thus unable to identify the exact species
which can pose another problem. It seems that the article only looked at
bacteria and not at fungi or perhaps even viruses, as these might have a role as
well. More on the limitations of 16s rRNA sequencing
What is important to remember, is that the age of
the samples in the first study can also have affected the results in other
ways. Both treatment of CF and antibiotic use aren't the same nowadays as they were in 1992. While the first study does measure whether or not the patients used antibiotics it did not specify what kind of antibiotics .
Another large limitation of these studies and of studies
with lung microbiota in general is that it is very hard to differentiate
between actual lung microbiota and microbes introduced from the upper airways
during bronchoscopy. Therefore it is possible that a lot of samples have been contaminated with upper airway bacteria. The second study to it's credit, does provide a detailed description of methods
used to prevent upper airway contamination. (2) It is unclear whether the same care has been applied in the first study.
Study results
The first article claims a dominance by so called recognized CF
bacterial pathogens like s. aureus, and as a consequence less diversity of the microbiota, was
associated with more inflammation, characterized by higher levels of neutrophil electalase but not IL-8.(1)
It also claimed that symptomatic children had less diversity in their microbiota. It could be argued however that this is due to symptomatic children frequently being treated with antibiotics. The article also claims symptomatic children usually have a higher level of inflammation. Thus, it can be argued that the reduced diversity in microbiome is not directly related to inflammation but to antibiotic use due to inflammation. The article defines several bacteria
that are related to inflammation in case they develop into dominant species. However, these results could have been influenced by antibiotic use and storage conditions. While cultures were used in both studies, some bacteria might have an easier time growing under culture conditions than others do. Antibiotics use is less of a factor in the other study because sample
gathering and sequencing happened with a short period between them. Thus antibiotic treatment would be similar to current standards. (2)
Age has been associated in the first article with the
emergence of typical CF pathogenic bacteria and the development of reduced diversity of the
microbiota.(1)
This, however, might also be the result of cumulative antibiotic use since
continuous antibiotic use can lead to less diversity as well. The presence of
respiratory symptoms was associated with a reduced diversity of the microbiota, as described in the post about the lung microbiome. Reduced microbiota diversity however did not always lead to respiratory
symptoms. This is important since it is in fact suggestive for the antibiotic
hypothesis explained earlier; a reduced diversity does not automatically lead
to respiratory symptoms but respiratory symptoms (which often is a reason for
antibiotic use) are associated with reduced diversity. The other study can not
really say anything about the microbiota over time since it was cross
sectional.(2)
The following figure from the second study shows the
differences of the CF microbiome compared to that of healthy controls(2) The innermost circle represents different bacterial taxa. The inner rings are the controls and the outer rings the CF patients which each ring being one study subject. Darker shades of blue mean the taxa is relatively less common in this subject than in the mean normal value and darker shades of pink mean it is relatively more common.
 |
Reference: Renwick J, McNally P, John B, DeSantis T, Linnane B, et al. (2014) The Microbial Community of the Cystic Fibrosis Airway Is Disrupted in Early Life. PLOS ONE 9(12): e109798. https://doi.org/10.1371/journal.pone.0109798
|
My conclusion
In my opinion, the results from both studies are a reason
to further investigate the lung microbiota in relation to outcomes in patient with
CF.(1, 2)
The first study had the limitation that it could not include any healthy controls.(1)
The second study could compare the differences between healthy patients
and those with CF but could not discuss the change over time and was very
limited due to its small sample size.(2)
So, while these studies yield some interesting results, they are still very far from a definite insight in the role of the lung microbiota. In my opinion the ideal study would combine
design elements from both these studies.I would like to follow two large cohorts over time, one of CF
patients and one with matched healthy controls and
comparing the two trying to compensate for influencing factors like antibiotic
use if possible, for instance by only including patients in the control group that also frequently use antibiotics.
Written by Toon Zonneveld
References
1. Frayman KB, Armstrong DS, Carzino R,
Ferkol TW, Grimwood K, Storch GA, et al. The lower airway microbiota in early
cystic fibrosis lung disease: a longitudinal analysis. Thorax. 2017.
2. Renwick J,
McNally P, John B, DeSantis T, Linnane B, Murphy P, et al. The Microbial
Community of the Cystic Fibrosis Airway Is Disrupted in Early Life. PLOS ONE.
2014;9(12):e109798.
3. Quanjer PH,
Stanojevic S, Cole TJ, Baur X, Hall GL, Culver BH, et al. Multi-ethnic
reference values for spirometry for the 3–95-yr age range: the global lung
function 2012 equations. European Respiratory Journal. 2012;40(6):1324-43.
4. Miyamoto-Shinohara
Y, Sukenobe J, Imaizumi T, Nakahara T. Survival of freeze-dried bacteria. The
Journal of General and Applied Microbiology. 2008;54(1):9-24.
5. Choo JM, Leong LEX, Rogers GB. Sample storage conditions
significantly influence faecal microbiome profiles. 2015;5:16350.