Mastitis diagnostics and performance monitoring: a practical approach

In this paper a review is given of frequently used mastitis diagnostic methods in modern dairy practice. Methods used at the quarter, cow, herd and regional or national level are discussed, including their usability for performance monitoring in udder health. Future developments, such as systems in which milk-derived parameters are combined with modern analytical techniques, are discussed. It is concluded that, although much knowledge is available and science is still developing and much knowledge is available, it is not always fully exploited in practice.


INTroDuCTIoN
Worldwide, much effort is put into mastitis research, knowledge-transfer programmes, herd health advisery programmes, and cow-based ambulatory work. Because healthy udders are economically profitable (Halasa et al. 2007), lead to a better quality product (Ma et al. 2000), and better cow welfare (Ekman and Sandgren 2006); all these programmes have the intention of ultimately improving udder health. To be able to judge whether these activities are successful, performance has to be monitored. Many possibilities exist to monitor udder health performance, which are not always fully exploited. In this review, mastitis diagnostics that are essential in performance monitoring are discussed. Monitoring udder health performance is impossible without reliable and affordable diagnostic methods. Thus, there is a constant need to improve these methods, be it accuracy, cost price, or convenience. The most frequently used diagnostic methods are somatic cell counting (SCC) and bacteriological culturing (BC) of milk. Currently, methods such as measurement of N-acetyl-b-Dglucosaminidase (NAG-ase), lactate dehydrogenase (LDH), electrical conductivity (EC), and molecular methods such as polymerase chain reaction (PCR) technology are used less frequently. The latter has promising possibilities for future applications, especially in strain identification (Zadoks and Schukken 2006).

ClINICAl DIAGNoSIS
Mastitis diagnosis starts with visual observation. The role of the milker herein is undisputedly very important. It is generally known that large differences exist between farms in clinical mastitis diagnosis (Lam et al. 1993). This can have two causes -differences in farmers' definitions of clinical mastitis or differences in observing and identifying clinical cases. The latter is greatly improved by forestripping, which is also an important part of udder preparation (Reneau 2001). For a good clinical diagnosis, it is essential to be able to see abnormalities, for which you need sufficient light (at least 250 lux) underneath the udder, where teat cups are attached (Hulsen and Lam 2007). In the field of udder health, definitions agreed upon by the National Mastitis Council (1996) ought to be followed. If there is any detectable change in quarter and/ or any observable abnormality in the milk, the quarter is defined as having clinical mastitis. It is wise not to ignore clinical symptoms. Abnormal milk, even the smallest clot is a signal that something is wrong within that quarter. It should be noticed and action should be taken accordingly.

SomATIC Cell CouNT
SCCs have been known to be an important indicator of intramammary infections (IMI) for years ). An easy, cheap, and quick cow-side test to estimate SCC on farm is the California Mastitis Test (CMT). It is a semi-quantitative SCC measure by forming a stringy mass from the reagent (3% sodium lauryl sulphate) and DNA out of disrupted cells. Although the CMT test is not very complicated, it is not always executed correctly in practice. As a consequence, its usability is sometimes questioned. For practical guidelines the reader should refer to Hulsen and Lam (2007). Although the CMT has been proven to be valuable for over 40 years (Luedecke et al. 1967), results obtained from fresh cows are difficult to interpret and CMT should not be used to detect pathogens in milk from cows earlier than four days postcalving (Sargeant et al. 2001;Dingwell et al. 2003). The CMT can be valuable, for instance, to assess therapy success based on SCC estimation after treatment in clinically recovered cows, or to identify the affected quarter within a cow with elevated SCC. Direct SCC measurement, of course, is much more accurate, but more expensive and not always available cow-side. If SCC has to be measured in large numbers of samples, they are generally sent to a laboratory using high capacity cell counters based on the flow cytometry principle, such as the Fossomatic cell counter. This is a reliable way to measure SCC (Miller et al. 1986). Nowadays, tests such as Porta SCC and the DeLaval Cell Counter are available, which seem to be accurate on-farm methods of estimating SCC (Barratt et al. 2003). Interpretation of SCC is sometimes difficult, because it is a variable parameter that is influenced by many factors, such as diurnal variation (Olde Riekerink et al. 2007), stage of lactation, parity, and fraction of the milk sampled (Sarikaya and Bruckmaier 2006). The most important factor influencing SCC, however, is bacterial IMI (Dohoo and Meek 1982). Milk samples for SCC measurement should be taken immediately before milking, after removing three squirts of milk. Because of the variability of SCC, for monitoring infection dynamics longitudinal data are necessary, which should be based on multiple test results . In literature many different thresholds are used to differentiate 'healthy' from 'unhealthy' quarters or udders. The SCC in truly uninfected quarters does not exceed approximately 70,000 cells/ml, slightly increasing with age and lactation days (Schepers et al. 1997). Optimal sensitivity and specificity of SCC at a threshold of 200,000 cells/ml as indicator of presence of IMI are estimated at 73% and 86%, respectively (Dohoo and Leslie 1991). Thus, an operational threshold of practical value with minimal diagnostic error, 200,000 cells/ml, was proposed by Schukken et al. (2003).

BACTerIoloGICAl CulTurING
Bacteriological culturing can be executed at herd, as well as cow and quarter level, each with its own specific goal. Bacteriological culturing is most often used as a diagnostic tool to solve mastitis problems. Knowledge on the infectious status of mammary glands, however, can also be very helpful to prevent transmission of pathogens by diagnosing a reservoir at an early stage. Additionally, historical BC results give herd-based information that can be helpful in optimising the treatment of future mastitis cases. Bulk milk BC, used for diagnostic purposes has been reviewed by Jayarao et al. (2004). They described that environmental mastitis pathogens found in bulk milk may also be derived from non-specific contamination from cow skin, bedding, manure, or water. Contagious pathogens, such as Staphylococcus aureus and Streptococcus agalactiae generally do come from IMI, but due to dilution, latent infections, and intermittent shedding, sensitivity of a single bulk milk BC is low. Successive samples over a period of time, however, are considered as a good indicator for these infections in the herd (Jayarao et al. 2004). Sensitivity of BC of a single bulk milk sample for Mycoplasma spp. likewise, is not a perfect test (Biddle et al. 2003). In regions where the prevalence of Mycoplasma is high, however, routine testing for Mycoplasma spp. is recommended (Kirk et al. 1997).
To judge IMI at the cow level, composite milk samples have been used for a long time (Williams 1937). This relatively cheap method is still used frequently and can be an important source of information. Test characteristics of composite samples for detection of Strep. agalactiae (Dinsmore et al. 1991) and Mycoplasma spp. (Biddle et al. 2003) are good. Approximately 40% of quarters infected with Staph. aureus, however, will not be cultured from composite milk samples . In situations where one is interested in culture results of staphylococci or environmental streptococci, a selection of quarters for culturing, based on SCC or CMT, is probably more costeffective and reliable than the use of composite samples. At the quarter level, approximately 10-40% of milk samples from cases of clinical mastitis yield no growth. This may be due to several reasons (i.e., too few or no bacteria present or pathogens that require special culture techniques such as Mycoplasma spp.). In subclinical mastitis, latent infections or shedding cycles may also play a role (Sears et al. 1990). Through the years several methods, such as preculture incubation, preculture freezing, and increased plate inoculation volumes were tested trying to increase the recovery rate of pathogens. Increasing the inoculation volume to 0.1 ml significantly increased sensitivity (Lam et al. 1996), as did preincubation for four hours (Dinsmore et al. 1992). Schukken et al. (1989) found that freezing milk before culturing increased only the number of coagulase negative staphylococci found, whereas Dinsmore et al. (1992) showed a significantly higher positive overall culture rate. In general, these methods do increase the number of positive cultures, but it has to be realised that the growth Peer reVIewed promotion also affects possible contaminants in the milk. For clinical mastitis diagnosis, it is important that BC results are available as soon as possible to optimise treatment results, to save costs and to prevent the ineffective use of antibiotics. For that reason, several commercial on-farm culturing systems such as the Minnesota Easy Culture System II and the Petrifilm system became available during recent years. Both are user-friendly systems, applying milk with a sterile cotton swab or a plastic pipette, and have test characteristics (compared to traditional culture methods) that are good enough to make it a useful management tool. Reading results from the Petrifilm system, however, needs some more experience and ability than reading the Minnesota system (Godden et al. 2007).

SeleCTING CowS For BACTerIoloGICAl CulTurING
To effectively use bacteriological culturing as a diagnostic tool, milk samples have to be collected from the correct cows and quarters at the correct point in time.
As discussed above, most information is gathered if samples are collected at the quarter level. The most practical approach to select cows is to use cow-level SCC information. To select quarters, either CMT or quarterlevel SCC can be used. SCC patterns are, to some extent, related to the type of mastitis causing pathogen (de Haas et al. 2004). Some IMI are of short duration, and thus of limited importance to the farmer. On the other hand, subclinical infections can have shedding patterns and thus may not always be diagnosed in a single BC sample (Sears et al. 1990). Ideally, subsequent quarter samples are collected to increase diagnostic properties. However, that approach is time consuming and expensive, and therefore not very practical. Comparing bacteriological data from single samples to the actual situation of those quarters based on subsequent samples (gold standard) in five dairy herds showed that predictive values of a single test increased when SCC information was used to select cows (Lam and Schukken 2005). The probability that a Staph. aureus infected cow will still be infected a month after initial sampling is twice as high in a cow that has had a SCC > 200,000 cells/ml repeatedly, compared with all cows with SCC > 200,000 cells/ml (Table 1). Thus, for approaching mastitis problems at the herd level, it seems wise to select cows with at least two consecutive samples with elevated SCC, measured on a monthly basis. In addition to SCC information, measuring NAG-ase can be informative in selecting samples for BC. In a study of 10 commercial dairy herds, Berning and Shook (1992) showed that, after selection of infected quarters based on SCC, the log NAG-ase was more effective in identifying major from minor pathogen infections. Thus, information on NAG-ase may be helpful in selecting IMI with major pathogens.

ClINICAl mASTITIS IN relATIoN To INTrAmAmmAry INFeCTIoNS
Clinical mastitis is the most costly form of mastitis, due to discarded milk, treatment, and other costs (Halasa et al. 2007). Subclinical mastitis leads to production losses, can be a source of new infections (Lam et al. 1996a), and is also related to the occurrence of clinical mastitis cases. In a study in seven low bulk milk SCC herds, IMI status of all quarters, as well as clinical mastitis cases were monitored in detail during a period of 20 months (Lam 1996). In these herds, approximately 40% of IMI never showed clinical signs (Table 2). Conversely, approximately 50% of clinical mastitis cases in these herds were part of chronic IMI, indicating that, even in herds with low SCC, clinical mastitis is to some extent associated with subclinical IMI. It is likely that this relation is even stronger in herds with high SCC.

moleCulAr meTHoDS For DIAGNoSTIC PurPoSeS
The use of molecular methods in pathogen detection has increased over the last years. Often, these methods use polymerase chain reaction (PCR) technology. Testing the presence of a specific bacterial species, a part of the DNA of that pathogen is amplified and subsequently visualised. For a number of mastitis pathogens, PCR-based techniques have been described (Lee et al. 1998;Baird et al. 1999;Hassan et al. 2001;Daly et al. 2002). These methods are currently very labour-intensive and it is expensive to do a separate PCR test for every possible mastitis pathogen. For that reason, multiplex PCR tests are of interest, in which several pathogens can be tested at the same time (Phuektes et al. 2003;Bottero et al. 2004). Additionally, real-time PCR assays are being developed (Lightcycler, Taqman, Luminex, Biacore) for detection and quantifying mastitis pathogens in milk. Molecular methods can also be used to differentiate bacterial strains within one bacterial species. These differences may be of importance, because they may be associated with differences in virulence, epidemiology, and cure rates. For these purposes, phenotypic characteristics,

All cows
Cows with SCC >200,000 cells/ ml  such as phage types, serotypes, and antibiotic sensitivity patterns can be used. Another possibility is to test for differences in strains by genotyping (fingerprinting) their genome. In epidemiological research, these methods are frequently used (Zadoks and Schukken 2006). Several molecular methods, such as PFGE (pulse field gel electrophoresis), ribotyping, RAPD (random amplified polymorphic DNA), AFLP (amplified fragment length polymorphism), and MLST (multi locus sequence typing) are used for genotyping. Generally, these methods use DNA that became available after digestion with restriction enzymes, after amplification with PCR techniques, DNA sequence analyses, or combinations of these techniques. For a review of these techniques, see Zadoks and Schukken (2006). In most countries PCR technology is not available for routine use in mastitis diagnostics. Recently more information became available on a commercial multiplex PCR test performed directly on milk, which is available for routine use (Koskinen et al. 2008).

FuTure DeVeloPmeNTS
Numerous different techniques to diagnose mastitis have been developed over the years. Electrical conductivity of milk is not a new method, but so far, using the test by itself, its characteristics, especially specificity, are not that good (Nielen et al. 1992), and its use on farm has been limited. New developments, such as the use of fuzzy logic (de Mol and Woldt 2001), and an approach through Bayesian Networks (Steeneveld et al. 2008) enhance test characteristics. To improve these, information from other sources is necessary to distinguish healthy cows, cows with subclinical, and cows with clinical mastitis (Norberg et al. 2004). Combining different types of milk-derived data, like yield, temperature and EC of milk, can be used for automated cow status monitoring. This can overcome problems like the occurrence of many false-positive alerts, when EC is used as a single parameter (Nielen et al. 1992). The inclusion of additional information like LDH, but also herd and udder characteristics, leads to early diagnosis of clinical mastitis cases (Chagunda et al. 2006). This methodology has been developed and is currently being tested in a number of commercial dairy farms in Denmark (Blom and Nielsen 2008). If it actually proves to be able to predict the occurrence of clinical mastitis a few days ahead (Friggens et al. 2007), it may be a step forward in mastitis diagnosis.

PerFormANCe moNITorING
The diagnostic methods discussed above, can be used to monitor udder health performance at the cow and the herd level, as well as the regional or national level.
To judge treatment outcome, the level of judgement is the quarter or the cow. In research, generally BC is used for this purpose, sometimes supplemented with SCC. Although they are of great importance, cure-rates are hardly evaluated quantitatively in daily practice. Because information concerning cure rates gives the possibility to really judge treatment success in a herd, the use of these data by farmers and practitioners should be advocated. Ideally, all clinical mastitis cases are sampled and results are used to optimise treatment of individual cows. If that is not practised, it is still recommended to collect samples of all cases of clinical mastitis, and store them (correctly identified) in a freezer. They can, if required, be cultured later. A general recommendation is to make up the bacteriological udder health status of a herd at least once a year, or more frequent if needed. For such a bacteriological udder health status, a selection of at least 10 high SCC cows and 10 clinical mastitis cows should be cultured using quarter milk samples (Lam et al. 2008). Based on the bacteriological information, combined with the evaluation of recent treatments, a standard treatment protocol for the forthcoming period can then be developed. At the herd level, udder health performance can be judged based on information on clinical mastitis and SCC, preferably supplemented with BC. This information is even more important than data on cure rates, because it is highly correlated with general udder health management. Based on the results of the previous year, realistic goals for SCC can be set for the next year, followed by an action plan, including a timeline. Estimates of the annual economic impact of mastitis for the herd can be used to motivate the farmer and prioritize actions. Performance of knowledge-transfer projects can also be monitored. The easiest approach is to measure bulk milk SCC, but clinical mastitis incidence at herd level and the effect on the use of antibiotics can also become available (Green et al. 2007). Finally, udder health performance can be measured on a regional or a national basis. Again, bulk milk SCC is the easiest parameter, which is monitored in many countries. Clinical mastitis incidence, antibiotic use, BC, and antibiotic sensitivity patterns of cultured bacteria, however, are much more informative than SCC data. In Scandinavian countries especially, valuable data on this subject are available and used, trying to further improve udder health in these countries (Østerås et al. 2007).

CoNCluSIoNS
Much knowledge in the field of mastitis diagnostics is available and science is still developing. Several diagnostic methods are available to monitor the effect of interventions in udder health, but these are not always fully exploited. At the cow level, cure rate after therapy is measurable, but hardly quantified. Clinical diagnosis, together with SCC measurement and BC are the most frequently used parameters for this aim. At the herd level, goal setting based on available retrospective data could be practised more often. The same parameters can be used, supplemented with SCC and bulk milk BC. Future developments, such as systems in which several milkobtained parameters such as EC and LDH are combined, may lead to changes in this field in the near future. At the regional or national level, only limited data are presented in scientific publications, whereas these data are often available and informative on the effect of certain Peer reVIewed approaches chosen in a certain area.

ACKNowleDGemeNTS
This article reports on a study which is part of the five-year mastitis programme of the Dutch Udder Health Centre and was financially supported by the Dutch Dairy Board.