Large-scale immunohistochemical examination for lymphoreticular prion protein in tonsil specimens collected in Britain


  • Mar Fernandez de Marco,

    1. Division of Neuropathology and Department of Neurodegenerative Disease, UCL Institute of Neurology, Queen Square, London WC1N 3BG, UK
    Search for more papers by this author
  • Jacqueline Linehan,

    1. MRC Prion Unit, Department of Neurodegenerative Disease, UCL Institute of Neurology, Queen Square, London WC1N 3BG, UK
    Search for more papers by this author
  • O Noel Gill,

    1. Centre for Infections, Health Protection Agency, 61 Colindale Avenue, London NW9 5EQ, UK
    Search for more papers by this author
  • Jonathan P Clewley,

    Corresponding author
    1. Centre for Infections, Health Protection Agency, 61 Colindale Avenue, London NW9 5EQ, UK
    • (Regarding the tonsil archive) Centre for Infections, Health Protection Agency, 61 Colindale Avenue, London NW9 5EQ, UK
    Search for more papers by this author
  • Sebastian Brandner

    Corresponding author
    1. Division of Neuropathology and Department of Neurodegenerative Disease, UCL Institute of Neurology, Queen Square, London WC1N 3BG, UK
    • (Regarding the IHC) Division of Neuropathology and Department of Neurodegenerative Disease, UCL Institute of Neurology, Queen Square, London WC1N 3BG, UK
    Search for more papers by this author

  • No conflicts of interest were declared.


There have been 173 cases of variant Creutzfeldt–Jakob disease (vCJD) in the UK, as of 5 July 2010, as a result of the bovine spongiform encephalopathy epidemic. The number of individuals subclinically infected with vCJD, and thus the eventual number of cases, remains, however, uncertain. In an attempt to address this problem, 63 007 tonsil tissue specimens were previously tested by enzyme immunoassay (EIA) for the presence of disease-related prion protein (PrPres) and found to be negative. To confirm the reliability of this result, all those in the birth cohort most at risk (1961–1985) and a few others, including controls, have now been tested by immunohistochemistry (IHC). Histological slides were prepared from 10 075 anonymized formalin-fixed, paraffin-embedded tissues and examined for PrPres with two anti-prion protein antibodies, ICMS35 and KG9. One specimen showed a single strongly positive follicle with both antibodies, on two slides from adjacent sections. As this specimen was negative when it was further investigated by EIA, IHC, and immunoblotting, it is unclear whether the patient from whom the tonsil came will go on to develop vCJD. If, however, this is the case, then a finding of 1 out of 9160 gives a prevalence of disease-related prion protein in the British population of 109 per million, with a 95% confidence interval (CI) of 3–608 per million, which is not statistically different (exact p = 0.63) from population prevalence estimates based on finding three positives out of 10 278 in a previous IHC study of appendix tissue. If this is not the case, a finding of 0 out of 9160 gives a prevalence of 0–403 per million (95% CI) for the 1961-1985 cohort, which is also not different (exact p = 0.25) from previous population prevalence estimates. Therefore, the results of this work could be summarized as finding, by IHC, no or one vCJD-positive individual. Copyright © 2010 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.


Variant Creutzfeldt–Jakob disease (vCJD) is understood to have arisen from bovine spongiform encephalopathy (BSE) 1–3. There was widespread population exposure in the UK and some other countries to BSE and as of 5 July 2010, at least 220 people have developed clinical vCJD worldwide (173 in the UK) 4. The number of currently subclinically infected individuals, and thus the eventual number of cases, remains uncertain 4–6. This represents an ongoing public health concern with the risk of iatrogenic transmission through blood and surgical instruments 7, 8, since prions resist most conventional decontamination procedures 9. Four instances of vCJD infection resulting from blood transfusion have been reported, establishing the existence of an infective asymptomatic stage 10–14. There has been a report of autopsy finding of abnormal prion protein (PrPres) in the spleen of a person with haemophilia 15. Iatrogenic transmission of sporadic CJD has also been reported to occur through neurosurgical instruments, and experimental studies have shown stainless steel-bound prions to transmit disease with remarkable efficiency when implanted into mice 16–18. These factors, together with the unknown maximum length of the asymptomatic incubation period and the influence of the host's genotype 19–21, all contribute to the uncertainty about the underlying prevalence of vCJD.

Preclinical colonization of the lymphoreticular system in vCJD is lent support by the detection of PrPres in an appendix removed 8 months before onset of overt neurological symptoms in a patient whose diagnosis was confirmed at autopsy 22, 23. The finding of PrPres in the spleen removed at autopsy from a person with haemophilia is consistent with lymphatic spread of disease 15, as is the report of PrPres, but not disease, in spleen and lymph tissue at post-mortem from a recipient of red blood cells donated by a vCJD case approximately 18 months before onset of clinical symptoms 12. Moreover, tonsil biopsies are successfully used for the diagnosis of vCJD, showing 100% sensitivity and specificity 24, 25. These collective data indicate that large-scale screening of surgical tonsillectomy tissues for PrPCJD could provide early warning of a high level of subclinical prevalence of vCJD prion in the general population 6, 22, 24, 26, 27.

Three previous studies analysed appendix and tonsil specimens for the presence of PrPres6, 26, 27. In the first study, 11 247 archived fixed appendix specimens, and 1427 tonsil specimens, were screened by immunohistochemistry (IHC), revealing PrPres deposition in three appendix specimens, all from the 1961–1985 birth cohort 26. The prevalence of detectable PrPres in Britain was therefore calculated to be 292 (95% confidence interval 60–853) per million 26. In a second study, 2000 tonsils were screened by both immunoblotting (IB) and IHC, showing no positive cases 27. A third study examined 63 007 tonsil specimens from a national anonymous tissue archive and screened for the presence of the PrPres by the use of two enzyme immunoassays (EIAs) based on different analytical principles 6. No samples contained detectable levels, allowing a prevalence estimate of 0 per million (upper 95% confidence limit of 113 per million). These results suggest that the prevalence of subclinical vCJD infection in Britain may be lower than, but still consistent with, that given by the survey of appendix tissue 26, with an upper limit in tonsil tissue of 289 per million in the 1961–1985 birth cohort 6. These two surveys may not, however, be directly comparable, particularly because the study of Hilton et al26 screened all of the samples by IHC, whereas the study of Clewley et al6 used EIA as the screening method and only employed IHC as a confirmatory method on a limited subset of the 63 007 tonsils.

The aim of the present study was to investigate further, by IHC, the prevalence of subclinical vCJD in Britain in anonymized samples derived from patients in the 1961–1995 birth cohort of the 63 007 tonsils collected by Clewley et al6.

Materials and methods

Tonsil archive

The tonsils used in this study came from an opportunistic sample of 63 007 tonsils removed for clinical reasons at 131 hospitals across England and Scotland representing 11 Strategic Health Authorities (South West, South Central, South East Coast, East of England, London, East Midlands, West Midlands, Yorkshire and the Humber, North East, North West, and Scotland) 6. All available tonsils from patients in the 1961–1985 birth cohort were selected for testing by IHC, as well as others chosen for control and technical reasons. The study received ethical approval from the Trent Multi-Centre Research Ethics Committee (MREC/03/4/073) 6.

Sectioning and conventional staining

Sections were prepared from tissue blocks at a nominal thickness of 5 µm at two levels using a standard rotary microtome (LEICA RM 2135, LEICA, Milton Keynes, UK). One section per block was stained with H&E for morphological assessment using standard staining procedures, on a LEICA Autostainer ST5020.

Immunohistochemical staining

For immunohistochemical analysis, sections of formalin-fixed tonsils were dewaxed and rehydrated, immersed into 98% formic acid for 5 min, and then washed in PBS. Thereafter, the slides were loaded onto a BondMax automated immunostaining instrument (LEICA). All antigen retrieval, staining, washing, and haematoxylin counterstaining steps were carried out on this instrument. The antigen retrieval was performed using Bond Epitope Retrieval solution 1 and Bond Enzyme (LEICA Microsystems, Milton Keynes, UK). Endogenous peroxidase was neutralized and the sections were incubated with either anti-PrP monoclonal antibody ICSM35 (D-Gen, London, UK; dilution 1 : 1500) or anti-PrP monoclonal antibody KG9 (Institute for Animal Health, TSE Resource Centre, UK; dilution 1 : 3000). ICSM35 recognizes the region encompassing residues 93 and 102 of human PrP, and KG9 recognizes residues 140–180 of human PrP. The sections were then incubated with the secondary antibody for signal amplification and detection (Vision Biosystems Bond Polymer Detection System, visualized with diaminobenzidine and Bond DAB Enhancer). After counterstaining with haematoxylin, the sections were dehydrated in ascending concentrations of alcohols and xylene and coverslipped with a LEICA ST5020 automated coverslipper.

Autopsy brain tissues from confirmed cases of CJD were used as a positive control for each machine cycle, while omission of the primary antibody on a CJD brain section served as a negative control. Blinded positive control sheep scrapie tonsil tissue cases were randomly included among the human tonsil tissue blocks as an internal quality control, to test the overall sensitivity of the IHC screening.

Each section was labelled with the unique identifier number, the date of the run, and a unique identifier that can be linked to all data related to the machine cycle, reagent and batch number, incubation times, and other parameters.

Microscopic examination

A first quality control was performed by evaluating the quality of the staining in the controls and in the tonsil specimens. We examined and scored entire sections of every anonymous tonsil specimen, at one or more levels, after inspection of a minimum of 15 lymphoid follicles or a minimum of 20 mm2 tonsil area.

Additional testing

EIA screening, immunoblotting (IB), and codon 129 genotyping were carried out as previously described 6. For further investigatory IHC, the existing and new wax blocks were sent to two independent laboratories (CJD Surveillance Unit, Edinburgh and Derriford Hospital, Plymouth, UK), where they were tested as previously described 6, 26. Confirmatory enhanced chemiluminescent IB was carried out by the MRC Prion Unit 25, 27 and the National CJD Surveillance Unit 11, 28.

Statistical methods

95% confidence intervals for prevalence estimates were calculated using the exact binomial method, and comparisons of prevalence between surveys were made using Fisher's exact test.


We examined a total of 24 360 slides by IHC, of which 17% were repeated because of failure of IHC or insufficient control staining (11%), enzymic overdigestion, irregular distribution of DAB, poor sectioning, etc (5%). One per cent of the slides were repeated for further investigation, due to the presence of suspicious staining. From 10 075 tonsil specimens, 4% were rejected due to the absence or too small an amount of lymphoid tissue, or because of poor tissue quality. Of the 9675 tonsil specimens accepted for the study, 94% had more than 30 lymphatic follicles, 5.5% contained 15–30 follicles, and only 0.5% had a minimum diagnosable area. We screened about 20% of anonymous tonsil specimens more than once. We found three samples with positive labelling requiring further examination to confirm or exclude specific labelling for PrPres.

The first specimen, 38 660 (Figures 1A–1D, and Table 1), when stained with ICSM35 (Figure 1A) and KG9 (Figure 1B), showed intense but diffuse staining in an identical follicle. However, while immunoreactivity was not seen elsewhere in the section stained with KG9, we could detect non-specific staining when the specimen was stained with ICSM35. Therefore, new sections were stained with ICSM35 or KG9 (Figures 1C and 1D), showing no positive staining in the follicle that was positive before. These slides were independently examined, with the conclusion that although the pattern of follicular dendritic cell (FDC) staining in this case does not have the coarse granularity seen in positive tonsils from symptomatic cases, it should be classed as positive on the basis of the one strongly positive follicle with both KG9 and ICSM35 antibodies (J Ironside and D Hilton, personal communication).

Figure 1.

Tonsil specimens with positive immunolabelling. (A–D) Immunoreactivity in tonsil specimen 38660 stained with ICSM35 (A), KG9 (B), ICSM35—new sections (C), and KG9—new sections (D). (E–H) Non-specific immunoreactivity in tonsil specimen 18824 stained with anti-PrP antibody ICSM35 (E), anti-PrP antibody KG9 (F), ICSM35—first repeat (G), and ICSM35—second repeat (H). (I–N) Immunoreactivity in tonsil specimen 40751 stained with ICSM35 (I), KG9 (L), ICSM35—first repeat (J, M), ICSM35—second repeat (K), and KG9—repeat (N). Scale bar: 320 µm (all images)

Table 1. Human tonsil tissues reactive by IHC in the 1981–1985 birth cohort
ID NoFiguresICSM35KG9Interpretation of IHCIBEIA resultPRNP codon 129
  • *

    By four independent methods: Bio-Rad, Prionics, and two ‘in-house’ methods.

  • By two independent methods: Bio-Rad and Microsens.

386601A– 1DInitial reactivityInitial reactivityOne strongly positive follicleNeg*NegMV
407511E– 1HInitial reactivityInitial reactivityProbable background stainingNeg*NegMV
188241I– 1NInitial reactivityInitial reactivityBackground stainingNeg*NegVV

A second specimen, 18 824 (Figures 1E–1H, and Table 1), showed strong but diffuse, rather than specific FDC staining. This tonsil showed three adjacent follicles positive in one margin of the specimen when stained with ICSM35 (Figure 1E), together with non-specific staining elsewhere in the tonsil. The same follicles were positive in the KG9 staining in an adjacent section (Figure 1F). Because the staining was suspicious but not typical, the staining was repeated on an adjacent section on a different instrument using antibody ICSM35 and established protocols (dilution 1 : 3000). Still there was speckled positive labelling of unclear significance in the three follicles that were positive before and no staining elsewhere (Figure 1G). On the basis of this result, new sections were prepared and stained, showing perinuclear positivity in the same (immediately adjacent) follicles (Figure 1H). These slides were independently examined, with the conclusion that the staining seen was ‘background’ (J Ironside and D Hilton, personal communication).

A third specimen, 40 751 (Figures 1I–1N, and Table 1), was stained with ICSM35 (Figure 1I) and KG9 (Figure 1L). Although no immunoreactivity was found in both slides, ICSM35 staining was repeated because of poor staining quality. Therefore, a new section was stained using ICSM35. Immunoreactivity was then detected in one lymphoid follicle, showing a fine granular pattern suggesting FDC positivity (Figures 1J and 1M). We therefore repeated the staining again on new sections, again with ICSM35 (Figure 1K) and KG9 (Figure 1N), respectively. This (second) repeat showed no immunoreactivity. These slides were independently examined, with the conclusion that the staining seen was ‘probable background’ (J Ironside and D Hilton, personal communication).

These three samples were further investigated by EIA, IB, IHC, and codon 129 genotyping (Table 2). They had all given negative results in the initial EIA screening, and this was confirmed on repeat testing after the IHC findings were reported. IB by both the Prionics (G Mallinson, personal communication) and the Bio-Rad methods was negative for all three samples. In addition, multiple tissue homogenates were referred as blinded samples for enhanced chemiluminescent IB 11, 25, 27, 28 and were reported as negative (J Wadsworth and M Head, personal communication). Additional tissue blocks in wax were made from each of these three samples and they were independently examined and reported as negative (J Ironside and D Hilton, personal communication).

Table 2. Prevalence of disease-associated prion protein (PrPCJD) in Britain: positive/total and rate per million with 95% confidence intervals
  Birth cohort
Survey 1961–1985All tested
  • *


  • Rate per million with 95% CIs.

  • NA = not applicable, as the 95% CI is calculated only when the denominator exceeds 1000.

2004–September 2008 national tonsil survey by IHC 1/9160*1/9672
Interpretation: one positive patient 109 (3–608)103 (3–576)
2004–September 2008 national tonsil survey by IHC 0/91600/9672
Interpretation: zero positive patients 0 (0–403)0 (0–378)
2004–September 2008 national tonsil survey by EIA 0/12 7530/63 007
  0 (0–289)0 (0–59)
1995–1999 national tissue survey by IHCAppendices3/10 2783/11 247
  292 (60–853)267 (55–779)
  NA0 (0–2582)

Tonsils 38 660 and 40 751 were both MV heterozygotes, whereas tonsil 18 824 originated from a patient homozygous for valine at codon 129 of the PRNP gene. All three samples were from patients in the 1981–1985 birth cohort.

Three internal quality control sheep scrapie tonsil tissues were successfully detected, although only when stained with anti-PrP monoclonal antibody ICSM35. None of these tonsils showed immunoreactivity when stained with anti-PrP antibody KG9 (Figures 2G–2I and Table 1). The sheep tonsils showed a different morphology to that of human tonsils. The first positive control sample showed immunoreactivity with ICSM35 in most of the follicles, with a fine granular pattern in a distribution compatible with FDCs, including also some coarse granular aggregates (Figure 2A). The presence of granular staining outside the follicles was also detected. The second positive control sample showed a mixture of fine granular staining in cells with the morphology of FDCs, including coarse granular aggregates, and accumulation within the cytoplasm of macrophages in most of the lymphoid follicles when stained for ICSM35 (Figure 2B). The third positive control sample showed strong staining in a large area within most of the lymphoid follicles when stained with ICSM35, with a distribution suggesting that it was within FDCs (Figure 2C). These three control specimens were then additionally stained on a different instrument (Ventana Medical Systems) using ICSM35 and established protocols (primary antibody dilution 1 : 3000), confirming the positive result described previously (Figures 2D, 2E, and 2F, respectively).

Figure 2.

Immunoreactivity in the three blinded positive control sheep scrapie tonsil tissue specimens. Specimen 47540 stained with ICSM35 (A, D) or KG9 (G). Specimen 47542 stained with ICSM35 (B, E) or KG9 (H). Specimen 47544 stained with ICSM35 (C, F) or KG9 (I). Scale bar: 180 µm


Of the 9675 samples for which an IHC result was obtained, 9160 were in the 1961–1985 birth cohort. The remainder of the samples were selected for IHC because they showed some reactivity in the original serological screening of the 63 007 tonsils by EIA with Bio-Rad and Microsens kits 6. In addition, there were three positive controls (sheep scrapie) among the 9675 samples submitted for IHC. Three samples (18 864, 38 660, and 40 751) gave IHC results that needed to be investigated more fully. Two of these IHC results were concluded to be background staining by three experts, while for the third it was concluded that there was one strongly positive follicle with both KG9 and ICSM35 antibodies. This could not be confirmed by analysis of slides made from further tissue samples embedded in wax, neither could it be confirmed by IB. This result raises the question of the significance and interpretation of a single positive follicle among the thousands from several sections that were examined, particularly in the light of the failure of IB to confirm the presence of PrPCJD in the tissue. Further investigation of tissue from this specimen by bioassay or protein misfolding cyclic amplification (PMCA) was considered not to be worthwhile because bioassay is unlikely to be more sensitive than enhanced chemiluminescent IB tests 11, 25, 27, 28 and PMCA is insufficiently robust 29.

Our finding of one PrPres-positive follicle by IHC can be interpreted as showing that there is one individual in the 9160 samples from the 1961–1985 birth cohort who will go on to develop vCJD. Alternatively, if a single positive follicle is indicative of an insufficient amount of PrPres to spread and cause disease, the interpretation is that there is no one in the 9160 samples from the 1961–1985 birth cohort who will go on to develop vCJD. The decision between these two interpretations needs to be considered in the context of the relative sensitivities of the different tests that were used, and also in the context of the pathological significance of a small quantity of PrPres in a tonsil. Although all three methods (EIA, IB, and IHC) are based on the recognition of PrPres by specific anti-PrP antibodies, they are qualitatively and quantitatively different. As just a few stained cells can be seen by IHC, it could be argued that it is the more sensitive technique. Conversely, however, as a greater volume of tissue and therefore a larger number of cells can be tested by EIA and IB, it can be argued that they are the more sensitive methods 15. However, the distribution of PrPres in the tissue is likely to be an important factor in assessing the comparative sensitivities of different tests: when there is a very focal deposition of PrPres, IHC may be assumed to have the advantage.

Therefore, while we cannot say whether the patient from whom this tissue came will go on to develop vCJD, we can be reasonably certain, however, that the patient has not yet developed disease as the codon 129 PRNP genotype is MV, and all probable and definite vCJD cases to date have been MM at this loci. There have been four ‘possible’ cases of clinical vCJD, one of which was MV, but this was not biochemically confirmed and it was in a different birth cohort from the person from whom the tonsil in our study came 30. Also, the two IHC positives (out of three) from the previous study 26 for which a codon 129 genotype could be determined were PRNP codon 129VV 31 and no vCJD cases of this genotype have been reported.

The prevalence in the British population of underlying disease-related prion protein calculated from these findings is, if specimen 38 660 came from a vCJD-positive person, 109 per million for the 1961–1985 birth cohort, with a 95% confidence interval (CI) of 3–608 per million (Table 2), which is not different (exact p = 0.63) to the finding of three positives from 10 278 samples for the appendix survey 26. If tonsil 38 660 did not come from a vCJD-positive person, then the prevalence is 0 per million with an upper 95% CI of 403 for the 1961–1985 cohort and 0 per million for the 1961–1995 cohort with an upper 95% CI of 394 (Table 2), which is not different (exact p = 0.25) from the previous study.

It is possible that infection arising from exposure to BSE could cause more than one type of prion disease 32–34. Strains other than that resulting in vCJD, if they exist, may have markedly different pathogenesis, tissue distributions, and structural forms of PrPres. In addition, it is possible that genetic variability in the population may alter the pathogenesis of vCJD, in that the timing and rate of PrPres in appendix and tonsil tissues may differ between individuals. Indeed, genetic differences may even determine the extent of lymphoreticular pathogenesis 31.

Given that the collection of tonsils in our study has occurred later than the collection of appendix samples in the earlier appendix survey, it is conceivable that tonsils have been collected from infected individuals further into the incubation period than is the case for those individuals whose appendices were tested in the earlier survey 26. Moreover, should the incubation period for prion disease be considerably longer in people with different genotypes, uncertainty about the timing of the appearance of detectable PrPres in these will increase, with concomitant implications for the interpretation of results of PrPres prevalence surveys 6.

Animal experiments have shown that high infectivity, and even disease, can be present in the absence of detectable PrPres35. However, this observation cannot be generalized, as PrPres has always been detectable in the lymphoid tissues that have been tested from vCJD patients 6, 25, 28. Data from animal experiments also show ‘clearance’ of PrPres after inoculation 35, 36. Therefore, the PrPres found in the earlier survey of appendix tissue 26 may conceivably have been transient and eventually cleared without resulting in clinical disease, and therefore the result of the appendix survey result may not be replicable by the current tonsil survey 6.

Although, statistically, the vCJD prevalence estimates in this work do not differ significantly from those obtained by calculating from the previous Hilton study 26, qualitatively they suggest that prevalence estimates may be cautiously lowered. However, in an attempt to provide statistically significant evidence to demonstrate this, a large-scale IHC survey of recently collected appendix tissue specimens for the presence of PrPres is underway.


We are grateful to Dino diPrinzio, Chantelle Wilkinson, and Juliette Kaggwa for the preparation of slides and immunohistochemical staining. We also thank the staff of the national anonymous tissue archive (Kelly Vogliqi, Chris Kelly, Sally Hayes, Jahnavi Joshi, and Tom Turner) and the staff of the CJD Section, Centre for Infections (Carole Kelly and Caroline Lawson) for their contributions to this project. We thank Nick Andrews of the Statistics Unit, Centre for Infections for statistical calculations. We thank Mark W Head, Alexander H Peden, and James Ironside of the CJD Surveillance Unit; David Hilton and Philip Edwards of Derriford Hospital; Jonathan DF Wadsworth and John Collinge of the MRC Prion Unit; and Gary Mallinson and Antony Wilkes of NHS Blood T̈ransplant for confirmatory assays. We thank members of the Expert Advisory Group on the Laboratory Testing Strategy for Large Scale Abnormal Prion Prevalence Studies, chaired by Philip Minor, for discussions and recommendations regarding this work. This is an independent report commissioned and funded by the Policy Research Programme in the Department of Health, UK. The views expressed in the publication are those of the authors and not necessarily those of the Department of Health.


BSE   bovine spongiform encephalopathy

CI    statistical confidence interval

DAB   diaminobenzidine

EIA   enzyme immunoassay

FDC   follicular dendritic cell

H&E   haematoxylin and eosin

HPA   Health Protection Agency

IB   mmunoblotting (western blotting)

IHC   immunohistochemistry

PBS   phosphate buffered saline

PMCA  protein misfolding cyclic amplification

PRNP  gene encoding the prion protein

PrPres   disease-related prion protein, specifically the proteinase-K resistant core (it is also referred to in the literature as PrPSc and PrPCJD)

vCJD   variant Creutzfeldt–Jakob disease

Author contribution statement

MM, JL, and SB performed the IHC. JPC was responsible for the tonsil archive. ONG originally initiated the project. MM, SB, and JPC wrote the manuscript.