Вы находитесь на странице: 1из 5

Why should we measure bacterial load when treating

community-acquired pneumonia?
Grant Waterera and Jordi Rellob
School of Medicine and Pharmacology, Royal Perth Purpose of review
Hospital, University of Western Australia, Perth,
Western Australia, Australia and bCritical Care
We focus on a number of studies in the past 2 years that herald a dramatic shift in how
Department, Hospital Universitari Vall dHebron, Vall we treat patients with not just community-acquired pneumonia (CAP), but potentially all
dHebron Institut Recerca (VHIR), CIBERES,
Universitat Auto`noma de Barcelona, Barcelona, Spain
Recent findings
Correspondence to Jordi Rello, MD, PhD, UCI Area
General anexe 5 planta, Hospital Universitari Vall Recent studies report that high bacterial load, and specifically pneumococcal load in
dHebron, Passeig Vall dHebron, 119-129 08035 CAP, appears to be significantly associated with worse outcomes. These findings
Barcelona, Spain
Tel: +34 93 274 62 09; fax: +34 93 274 60 62; change the sepsis paradigm. Bacterial load may identify potential candidates for
e-mail: jrello.hj23.ics@gencat.cat adjunctive therapy, ICU admission and more aggressive management.
Current Opinion in Infectious Diseases 2011,
24:137141 Whereas we all acknowledge the importance of the virulence of the pathogen in the
outcome of CAP, microbiological tests currently play little role in management of
patients. Whereas molecular tests such as polymerase chain reaction have promised to
deliver accurate results in a clinically useful period of time, apart from a few niche
situations they have yet to enter routine practice. In particular the ability to calculate the
bacterial load in blood, and specifically pneumococcal load in CAP, appears to have
significant clinical utility. Not only does bacterial load predict clinical outcome, the data
so far available challenge some of our fundamental assumptions about optimal antibiotic
therapy and the pathogenesis of severe sepsis.

bacterial load, pneumonia, rt-PCR, sepsis, septic shock, Streptococcus pneumoniae

Curr Opin Infect Dis 24:137141

2011 Wolters Kluwer Health | Lippincott Williams & Wilkins

children over 30 years ago showed the amount of

Introduction Streptococcus pneumoniae or Haemophilus influenzae present
For more than a decade physicians have been comfor- was a strong predictor of clinical outcome, especially at
table using systemic viral load as a key therapeutic and levels above 103 organisms per ml [3,4]. Whereas calcu-
prognostic indicator in patients with a range of viral lating an organism load by performing quantitative blood
diseases such as human immunodeficiency virus (HIV), cultures is not difficult, it is time-consuming and there-
hepatitis C virus and cytomegalovirus. Intuitively it fore even less useful than conventional cultures in guid-
makes sense that the more pathogen there is in the ing clinical management.
patients blood, the worse the likely outcome. It therefore
seems surprising that it is only recently that the same
paradigm has been extended to bacterial infections. In Development of molecular assays including
this review we will outline recent research that is not only real-time PCR
offering new potential diagnostic and prognostic tools in The detection of bacteria by molecular assays, such as
patients with bacterial sepsis, and pneumonia in particu- polymerase chain reaction (PCR), has promised to
lar, but is also challenging some of our long-held beliefs replace conventional culture methods for over a decade.
about the pathogenesis of severe sepsis. However, apart from specific niche applications, newer
assays have failed to change our reliance on conventional
The concept that the greater the burden of bacteria in culture techniques. In pneumonia, developing a PCR
blood the worse the likely outcome for a patient is not assay for pneumococcus such that it has a sensitivity even
new. Quantitative blood culture studies in adults in the close to that of culture has been problematic [5]. One of
1930s and 1950s suggested that a higher burden of the major hurdles is the small reaction volume for PCR,
pneumococci in blood predicted a poor clinical outcome typically 0.2 ml, making it difficult to detect low levels of
[1,2]. Again studies using quantitative blood cultures in bacteremia that culturing 2040 ml of blood for 23 days
0951-7375 2011 Wolters Kluwer Health | Lippincott Williams & Wilkins DOI:10.1097/QCO.0b013e328343b70d

Copyright Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
138 Respiratory infections

will reveal. As the current gold standard of blood cul- advantage of a single comprehensive assay; however, until
tures is known to be imperfect, particularly in the setting such assays reach the sensitivity of pathogen-specific
of recent antibiotic exposure, it has also been hard to assays their utility remains questionable.
establish the specificity of PCR assays as the status of a
positive result in the absence of any other microbiological Whereas our lytA rt-PCR assay represents a significant
confirmation remains unclear. improvement in sensitivity over previous assays, as a
qualitative test it still largely fails the clinical so what
Thanks to improvements in the isolation of bacterial test in that, like pneumococcal urinary antigen, a positive
DNA from clinical samples and greater reaction volumes result is unlikely to change antibiotic selection. All
(1.0 ml), we were able to improve the performance of our appropriate guidelines recommend covering pneumo-
whole blood pneumococcal real-time PCR assay (rt-PCR) coccal infection empirically, so a positive result will
to a point when we felt it was clinically useful [6]. This not lead to any additional antibiotic coverage. The lytA
new pneumococcal assay targeted the autolysin gene rt-PCR, like the urinary antigen test, does not indicate
(lytA), had no cross-reactivity with other bacteria includ- antibiotic sensitivity so antibiotics chosen to cover resist-
ing other streptococci, and importantly there were no ance will not be changed. As there is also solid observa-
positive results seen in healthy controls [6], as occurs with tional evidence supporting a combination of antibiotics
pneumococcal urinary antigen assays [7]. Subsequent that includes a macrolide for bacteremic pneumococcal
analysis of 1800 samples taken as part of normal clinical pneumonia [12], a positive result for S. pneumoniae
practice at the same time as blood cultures in our Emer- will also not lead to a reduction in dual antibiotic
gency Department showed that the effective specificity therapy. Given the high frequency of multipathogen
of the lytA pneumococcal rt-PCR assay is between 99.5 disease in CAP in some series, a positive pneumococcal
and 100% [8]. result also does not rule out the presence of other patho-
gens, again limiting the likelihood of narrowing antibiotic
That we were able to achieve such a high specificity with coverage.
the lytA rt-PCR assay was not surprising as, unlike
respiratory tract specimens, pneumococci are not found The only potential impact on antibiotic therapy of the
in the blood of healthy individuals. The most significant qualitative lytA rt-PCR assay is in patients with severe
improvement over previous assays was the sensitivity of disease admitted to the intensive care unit (ICU), when
the lytA assay, with a yield of approximately twice that of much broader antibiotic coverage is often used (such as
blood cultures in a study of a cohort of 353 patients with carbapenems and vancomycin) in case multiresistant
community-acquired pneumonia (CAP) from a group of Staphylococcus aureus or uncommon Gram-negative patho-
hospitals in Spain more than double that previously gens resistant to third-generation cephalosporins or
reported [9]. Other recent studies using improved iso- fluroquinolones are the cause. In these severe cases
lation of bacterial DNA and pathogen-specific assays of pneumonia, a positive whole blood lytA rt-PCR would
have also demonstrated PCR yields of 150200% of that indicate that organisms resistant to usual CAP therapy are
of blood cultures [10]. One important implication of these extremely unlikely to be present, which should lead to a
recent studies is that minimum benchmark for any new reduction in the use of extremely broad antibiotic use
bacterial assays should be at least 150% of the blood that drives much of the antibiotic-resistant nosocomial
culture rate; otherwise it is likely that there are problems pathogen problems in ICUs.
with either bacterial DNA isolation or the assay itself that
need optimization. As the most sensitive assays are
currently detecting down to as low as 15 organisms/ The utility of assessment of quantitative load
ml, it is likely that further improvements in technology By establishing a bank of seeded control samples and
will increase the yield of assays even further. using the ability of the rt-PCR platform to calculate the
rate of amplification of the target DNA, the lytA assay can
With pneumonia we are fortunate in that S. pneumoniae is quantitate the amount of bacteria present in the original
by far the dominant pathogen; however, in sepsis more sample [6]. The current range of detection has varied
generally the range of bacteria is vast. Several approaches from as low as 3 cfu/ml through to 2.9 million cfu/ml,
have been tried to develop a single assay that would detect indicating the enormous range of bacterial burden in
all or most bacteria. However, it is clear that with current patients with pneumococcal bacteremia.
generation technology targeting a common bacterial gene
such as 16S or 24S does not have the same sensitivity as Analysis of the lytA rt-PCR calculated bacterial load and
pathogen-specific assays. Likewise multiplexing assays to clinical outcomes in the Spanish cohort study showed an
improve the range of bacteria detected also compromises extremely strong correlation with high bacterial loads and
sensitivity for individual pathogens [11]. Research will increased risk of both septic shock and mortality [9].
continue in these areas due to the obvious clinical Interestingly, across the entire cohort of 353 patients the

Copyright Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Bacterial load and community-acquired pneumonia Waterer and Rello 139

lowest mortality was seen in patients with a negative lytA A further potential for quantitative rt-PCR was demon-
result (2.8%), despite some of these cases having proven strated by Ho et al. [17]. In this study of 20 patients
bacteremia with other pathogens. There also appeared to with methicillin-resistant Staphylococcus aureus (MRSA)
be a threshold effect such that above 10 000 cfu/ml there bacteremia in an ICU in Taiwan, not only did nonsurvi-
was a marked increase in mortality (28.6 vs. 13.6%). An vors have much higher levels of MRSA DNA in whole
even more marked effect was seen on the risk of septic blood, but persistence of high DNA levels at day 3 and
shock with high bacterial loads, possibly because infec- day 5 was predictive of an adverse outcome. This
tion is clearly the cause of septic shock, whereas 30-day suggests that serial monitoring may have additional value
mortality is a more complicated outcome contributed to to assessment of bacterial load at presentation.
by many factors other than the acute infection [13].

Whereas the lytA rt-PCR findings need to be replicated Why should we measure bacterial load?
in other cohorts, there is already substantial support for Although further studies are needed, it is already appar-
the concept of rt-PCR estimated bacterial load predicting ent that bacterial load is a useful predictor of clinical
clinical outcome. Peters et al. [10] calculated bacterial outcome in patients with pneumococcal pneumonia. As
load in 45 patients with pneumonia using a lytA rt-PCR the assay has a current turn around time of around 2 h,
assay. A significant correlation between pneumococcal uses technology widely available in most hospital labora-
bacterial load in blood and the likelihood of the patient tories and currently has a consumable cost of under
developing systemic inflammatory response syndrome US$20 for reagents, it has significant potential to be
was observed, and similar to our findings Peters et al. incorporated into existing empiric guidelines for both
[10] also observed higher bacterial loads in patients who site-of-care decisions and empiric antibiotic selection.
were both PCR and blood culture positive compared to Whereas studies need to be done to see how bacterial
those who were only positive by PCR. load should modify existing scoring systems such as the
CURB-65 and the pneumonia severity index, it is clearly
Carrol et al. [14] studied a cohort of children in Malawi much more informative than tests routinely ordered such
with pneumococcal pneumonia and meningitis (n 82) as a white blood cell count, and certainly more clinically
or pneumonia (n 13). Bacterial load was calculated useful than blood cultures.
using a pneumolysin rt-PCR assay in both blood and
cerebrospinal fluid (CSF), with a strong correlation Perhaps more clinically important, what we have
observed with mortality and levels of interleukin (IL)-1 observed is that a high bacterial load at presentation is
beta and IL-10. Interestingly blood bacterial loads in this a very strong predictor of developing shock, even in
pediatric cohort were substantially lower than has been patients who are initially clinically stable. Existing scor-
observed in the adult cohorts. ing systems are very poor at picking patients who will
deteriorate after presentation, yet this group has been
In a cohort of 169 children with pneumococcal pneumo- identified as having particularly high mortality rates
nia, including 145 with empyema, Munoz-Almagro et al. [18].
[15] assessed pneumococcal bacterial load in plasma and/
or pleural fluid. Pneumococcal bacterial load was assessed As well as aiding the site of care decision initially, the
using a pneumolysin target. Both plasma and pleural fluid association between high bacterial loads and subsequent
bacterial loads were greater in patients with a hospital clinical deterioration raises some very important ques-
stay of 8 days or longer. In addition pleural bacterial load tions about fundamental assumptions we have had about
positively correlated with the length of pleural drainage. antibiotic therapy for pneumonia. What we hypothesize
As might be expected pneumococcal bacterial load was is that in patients with a high pneumococcal blood load,
much higher in pleural fluid than plasma. antibiotic-induced bacterial lysis plays a key role in
inducing sepsis. Whereas antibiotic-induced endotoxin
Bacterial load determined by rt-PCR also appears to be release and associated cytokine production has long been
useful in disease due to other pathogens. In a large study recognized as a contributing factor in patients with Gram-
of 1045 patients with proven Neisseria meningitidis infec- negative sepsis [19], this has been much more poorly
tion in the UK, Darton et al. [16] observed median documented in Gram-positive infections. In patients with
bacterial load from whole-blood rt-PCR was much higher pneumonia, bacteremia is associated with an excess of the
in patients who died (5.29 vs. 3.79 log10 copies/ml). sepsis-associated cytokines IL-6, IL-10 [20], and tumor
Complications such as limb or digit loss and hemodialysis necrosis factor (TNF) [21]. Interestingly, combination
were also predicted by higher bacterial load at presen- antibiotic therapy with a beta-lactam and a macrolide is
tation. Another key observation from this study was that also associated with a faster reduction in IL-6 than beta-
prior antibiotic administration did not appear to affect lactam monotherapy [22]. It is possible that one of the
bacterial load. reasons the mortality rate of pneumococcal bacteremia

Copyright Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
140 Respiratory infections

has remained relatively unchanged over the past 40 years J.R. has nothing to disclose and does not have conflict of interest. G.W.
[23], as well as the reason why antibiotics seem to make declares that he has nothing to disclose and has no conflicts of interest
related to the material presented in this manuscript.
little impact on mortality in the first 2448 h [24], is that
antibiotic therapy itself may initiate the chain of events
leading to early deaths in the small subset of patients with References and recommended reading
high bacterial loads. Papers of particular interest, published within the annual period of review, have
been highlighted as:
 of special interest
If the rapid lysis hypothesis is correct, we need to sub-  of outstanding interest
stantially re-evaluate the choice of antibiotics in patients Additional references related to this topic can also be found in the Current
World Literature section in this issue (p. 174).
with high blood pneumococcal bacterial loads. Not only
the choice of empiric therapy may be altered to select 1 Landsman JB. Bacteriaemia and prognosis in lobar pneumonia: the results of
quantitative blood culture in pneumococcus pneumonia. Glasgow Med J
antibiotics with less potential for rapid bacterial lysis, but 1952; 33:3345.
there is also evidence that the speed of administration 2 Christie IM. Bacteraemia in lobar pneumonia. Lancet 1933; 2:804805.
and dosing interval may impact on cytokine production 3 Santosham M, Moxon ER. Detection and quantitation of bacteremia in child-
[25]. Not only changes in antibiotic selection are likely. hood. J Pediatr 1977; 91:719721.
Since high bacterial loads predict a high risk of septic 4 Sullivan TD, LaScolea LJ Jr, Neter E. Relationship between the magnitude of
bacteremia in children and the clinical disease. Pediatrics 1982; 69:699
shock in patients currently stable, we now have a window 702.
of opportunity to exploit before sepsis is established. 5 Avni T, Mansur N, Leibovici L, Paul M. PCR using blood for diagnosis of
This is critical because all trials of antishock therapies invasive pneumococcal disease: systematic review and meta-analysis. J Clin
Microbiol 2010; 48:489496.
(such as anti-TNF antibodies, antiendotoxin antibodies,
6 Kee C, Palladino S, Kay I, et al. Feasibility of real-time polymerase chain
high-dose corticosteroids, etc.) have required the pre- reaction in whole blood to identify Streptococcus pneumoniae in patients with
sence of severe sepsis before initiation. We can expect community-acquired pneumonia. Diagn Microbiol Infect Dis 2008; 61:72
many of the therapies that had promising animal and/or 7 Gutierrez F, Masia M, Rodriguez JC, et al. Evaluation of the immunochromato-
early human studies but failed to show a benefit in phase graphic Binax NOW assay for detection of Streptococcus pneumoniae
III studies to be revisited now that we can clearly identify urinary antigen in a prospective study of community-acquired pneumonia in
Spain. Clin Infect Dis 2003; 36:286292.
a very high-risk group of patients prior to the onset of 8 Kee C, Fatovich DM, Palladino S, et al. Specificity of a quantitative real-time
organ failure. polymerase chain reaction assay for the detection of invasive pneumococcal
disease: identifying Streptococcus pneumoniae using quantitative polymer-
ase chain reaction. Chest 2009; 137:243244.
Finally, whereas data are not yet available in pneumonia, 9 Rello J, Lisboa T, Lujan M, et al. Severity of pneumococcal pneumonia
the study by Ho et al. [17] suggests that serial monitoring of  associated with genomic bacterial load. Chest 2009; 136:832840.
A novel study that changes the sepsis paradigm, emphasizing the importance of
bacterial load may be useful to customize the optimal bacterial load on outcomes.
duration of antibiotic therapy in patients with pneumonia. 10 Peters RP, de Boer RF, Schuurman T, et al. Streptococcus pneumoniae DNA
load in blood as a marker of infection in patients with community-acquired
pneumonia. J Clin Microbiol 2009; 47:33083312.
11 Azzari C, Moriondo M, Indolfi G, et al. Realtime PCR is more sensitive than
Conclusion multiplex PCR for diagnosis and serotyping in children with culture negative
Quantitative assessment of bacterial load in blood has pneumococcal invasive disease. PLoS One 2010; 5:e9282.
enormous potential to change the way we approach 12 Waterer GW, Rello J, Wunderink RG. Management of community-
 acquired pneumonia in adults. Am J Respir Crit Care Med 2010 [Epub
patients with pneumonia. Whereas the identification of ahead of print].
S. pneumoniae is of some clinical use, early data suggest A critical review on advances on CAP management.
that a high bacterial load predicts a high risk of adverse 13 Mortensen EM, Coley CM, Singer DE, et al. Causes of death for patients with
outcomes. Even more importantly, the presence of a high community-acquired pneumonia: results from the Pneumonia Patient Out-
comes Research Team cohort study. Arch Intern Med 2002; 162:1059
bacterial load can identify patients who will develop 1064.
severe sepsis prior to this being evident clinically, open- 14 Carrol ED, Guiver M, Nkhoma S, et al. High pneumococcal DNA loads are
ing a new window for interventions. If the early studies associated with mortality in Malawian children with invasive pneumococcal
disease. Pediatr Infect Dis J 2007; 26:416422.
are confirmed, very long standing beliefs about the
15 Munoz-Almagro C, Gala S, Selva L, et al. DNA bacterial load in children and
optimal therapy of bacterial infections will be overturned  adolescents with pneumococcal pneumonia and empyema. Eur J Clin Micro-
and there will need to be a major change in empiric biol Infect Dis 2010 [Epub ahead of print].
Study emphasizing the value of bacterial load in plasma for pneumococcal
guidelines not only in pneumonia, but in sepsis generally. pneumonia in children.
Finally it is possible that serial assessment of bacterial 16 Darton T, Guiver M, Naylor S, et al. Severity of meningococcal disease
load may be useful in monitoring patient progress and associated with genomic bacterial load. Clin Infect Dis 2009; 48:587594.
determining the duration of antibiotic therapy and 17 Ho YC, Chang SC, Lin SR, Wang WK. High levels of mecA DNA detected by
a quantitative real-time PCR assay are associated with mortality in patients
inpatient care. with methicillin-resistant Staphylococcus aureus bacteremia. J Clin Microbiol
2009; 47:14431451.
18 Restrepo MI, Mortensen EM, Rello J, et al. Late admission to the ICU in
Acknowledgements  patients with community-acquired pneumonia is associated with higher
Supported in part by AGAUR 2009/SGR/1226 and PCI Pneumonia- mortality. Chest 2010; 137:552557.
CIBERES. Patients with delayed admission are associated with worse outcome.

Copyright Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Bacterial load and community-acquired pneumonia Waterer and Rello 141

19 Holzheimer RG. Antibiotic induced endotoxin release and clinical sepsis: a 23 Rodriguez A, Lisboa T, Blot S, et al. Mortality in ICU patients with bacterial
review. J Chemother 2001; 13 (Spec No 1):159172.  community-acquired pneumonia: when antibiotics are not enough. Intensive
Care Med 2009; 35:430438.
20 Martinez R, Menendez R, Reyes S, et al. Factors associated with inflammatory
This study addresses the concept that antibiotics are not enough in manage-
cytokine patterns in community-acquired pneumonia. Eur Respir J 2010
ment of severe CAP and some patients may benefit from adjunctive
[Epub ahead of print].
21 Calbo E, Alsina M, Rodriguez-Carballeira M, et al. The impact of time on the
 systemic inflammatory response in pneumococcal pneumonia. Eur Respir J 24 Austrian R, Gold J. Pneumococcal bacteremia with special reference to
2010; 35:614618. bacteremic pneumococcal pneumonia. Ann Intern Med 1964; 60:759
Right therapy the first time should be a priority in management of pneumococcal 776.
25 Buijs J, Dofferhoff AS, Mouton JW, van der Meer JW. Continuous adminis-
22 Padrones S, Garcia-Vidal C, Fernandez-Serrano S, et al. Impact of antibiotic tration of PBP-2- and PBP-3-specific beta-lactams causes higher cytokine
therapy on systemic cytokine expression in pneumococcal pneumonia. Eur J responses in murine Pseudomonas aeruginosa and Escherichia coli sepsis.
Clin Microbiol Infect Dis 2010; 29:12431251. J Antimicrob Chemother 2007; 59:926933.

Copyright Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.