Immunoperoxidase versus immunofluorescence in the assessment of human renal biopsies

Immunoperoxidase Versus Immunofluorescence in the Assessment of Human Renal Biopsies Johan Mölne, MD, PhD, Michael E. Breimer, MD, PhD, and Christian T. Svalander, MD, PhD ● Background: For half a century, immunofluorescence (IF) on frozen sections has been the gold standard for immunohistochemical evaluation of renal biopsy specimens. In routine diagnostic immunohistopathologic evaluation, traditional IF has been replaced to a large extent by immunoperoxidase (IP) methods applied to paraffin sections of formaldehyde-fixed tissue. This is caused in part by the practical disadvantages inherent in the IF method, eg, separate tissue specimen and handling, UV microscopy, fading and impermanence of the label-making archiving, and difficult later investigation. Our aim for the present study is to evaluate IP as an alternative to IF in the diagnostic assessment of renal biopsy specimens. Methods: Proteolytic antigen retrieval, antibodies effective on deparaffinized sections, a sensitive detection system (Dako EnVision HRP; Dako, Copenhagen, Denmark), and a standardized and rigorously controlled procedure were applied to a series of renal biopsy specimens (n ⴝ 81) previously classified by means of light microscopy (LM) and IF. Staining for immunoglobulin G (IgG), IgA, IgM, C1q, and C3c were recorded as positive or negative for IF and IP in paired proportions, presuming that IF was the test standard. Results: Concordant observations were 71% for all (282 of 398 observations), 82% for IgG (65 of 79 observations), and 89% for IgA (72 of 81 observations). The majority of discordant observations (74 of 116 observations) were positive by means of IP, with mesangial deposits of IgM and C1q that were not found by IF. Statistically, there was no significant difference in outcomes between IF and IP for IgG, IgA, and C3c (P > 0.2). In addition, IP staining allowed simultaneous evaluation of tissue by LM and therefore correlation between tissue structure and immune deposits not readily attained by IF. Conclusion: In the present study, it is documented that for the detection of IgG, IgA, and C3c, IP applied to protease-digested deparaffinized sections of formaldehyde-fixed renal tissue is, with few exceptions, equal to IF on frozen sections. The EnVision HRP method used here is several times more effective in terms of primary antibody dilution than earlier existing IP methods, and because the avidin-biotin system is not involved, very little nonspecific background staining will occur. Discordant observations (116 of 398 observations; 29%) were in the majority (91 of 116 observations) due to positive IP findings of IgM and C1q, which deserve additional investigation. Am J Kidney Dis 45:674-683. © 2005 by the National Kidney Foundation, Inc. INDEX WORDS: Immunohistochemistry (IH); immunoperoxidase (IP); immunofluorescence (IF); renal biopsy.

I

MMUNOHISTOCHEMISTRY (IH) is an indispensable part of the histopathologic investigation of renal tissue. The current classification of glomerular diseases is based on light microscopy (LM) in combination with IH and electron microscopy.1,2 Not only the presence or absence of immunoglobulins, complement factors, and inflammatory agents, but also the presence of

From the Departments of Clinical Pathology and Surgery, Sahlgrenska University Hospital, Göteborg, Sweden. Received April 26, 2004; accepted in revised form December 20, 2004. Originally published online as doi:10.1053/j.ajkd.2004.12.019 on February 22, 2005. Supported in part by grants from the Swedish Medical Research Council, project no. 11621, and the L-E Gelin Memorial Foundation. Address reprint requests to Johan Mölne, MD, PhD, Department of Clinical Pathology, Sahlgrenska University Hospital, SE-413 45 Göteborg, Sweden. E-mail: johan. [email protected] © 2005 by the National Kidney Foundation, Inc. 0272-6386/05/4504-0007$30.00/0 doi:10.1053/j.ajkd.2004.12.019 674

various endogenous and exogenous antigens, is of importance to establish the diagnosis.3,4 The technique to detect cell and tissue antigens with fluorescein isocyanate–labeled antibodies, ie, immunofluorescence (IF), originally was invented and applied by Coons et al5 in the 1940s. IF was adopted early by renal pathologists and since the 1950s is incorporated in the diagnostic procedure for assessment of renal biopsy specimens. Since then, IF with 1 (direct) or several layers (indirect) applied to frozen tissue sections is the gold standard technique to evaluate renal biopsy specimens, and considerable knowledge has been gained.6 During the latest decades, there has been great progress in the technical development of immunochemical methods, especially regarding labeling and detection systems, giving greater efficiency and practicability.7,8 In addition, much experience has accumulated regarding the importance of proper histotechnical handling of tissues, ie, fixation, dehydration, embedding, sectioning, and so on to obtain reproducible results.9 Methods applicable to paraffin sections of form-

American Journal of Kidney Diseases, Vol 45, No 4 (April), 2005: pp 674-683

IMMUNOPEROXIDASE STAINING OF RENAL BIOPSIES

675

aldehyde-fixed tissue, omitting the need for frozen sections, have been developed. In addition, labeling with enzymes, eg, horseradish peroxidase (HRP) and phosphatase, gives a reaction product visible in plain light, omitting the need for fluorescence (UV light) detection.10 As a consequence, many initial difficulties, such as high background and spurious results, in using immunoperoxidase (IP) on deparaffinized sections have been eliminated or minimized. Various labeled and unlabeled immunoenzyme techniques, especially IP, are now applied routinely for IH in most areas of diagnostic histopathology.11 During the last 2 decades, vast numbers of observations have been published concerning every conceivable antigen detectable with enzyme-labeled IH methods.12 Regarding renal biopsies, several early reports showed immune deposits by applying IP methods to paraffin sections.13-16 Most important, the need for proteolytic digestion as an initial step before immunostaining of formaldehyde-fixed renal tissue was pointed out.17 Still, the traditional IF method is well established and commonly recommended,18 including a variant of the original IF method involving a separate washing/transport solution.19 However, IF has a main drawback in the need for processing another tissue sample in addition to the tissue used for LM and electron microscopy. Earlier attempts to use deparaffinized sections for IF20,21 and later modifications22,23 met with no apparent success despite encouraging results reported by these investigators. During several years, in our diagnostic work with renal biopsies, we have tested various IP methods on deparaffinized sections of formaldehyde-fixed tissue in parallel with IF. Our results indicate that IP is a very sensitive and reliable method for the demonstration of carbohydrate antigen, immunoglobulin, and complement deposits in renal biopsy specimens.24-27 IP has several practical advantages compared with IF and, in most cases, may replace IF as a routine method in the assessment of renal biopsy specimens. However, rigorous standardization and control of all steps in the process are imperative. The present report is a detailed account of our experience with a 2-step IP staining technique based on HRP-labeled dextran-conjugated antibodies28,29 applied to the diagnostic evaluation

of renal biopsy specimens. The result of IP on formaldehyde-fixed paraffin-processed tissue is compared with that of direct IF performed on frozen tissue from the same biopsy. Results show that IP on formaldehyde-fixed tissue samples can replace the standard IF technique. METHODS

Renal Tissue Samples The Department of Clinical Pathology and Cytology, Sahlgrenska University Hospital, Göteborg, Sweden, processes more than 500 renal biopsies (55% transplant and 45% native) annually for diagnostic investigation. Percutaneous core needle biopsy specimens are sampled from native (16-G needle) and transplanted kidneys (18-G) with a biopsy gun under ultrasound guidance. One core of renal tissue is placed in a moistened container on ice, and 1 or preferably 2 cores are immersed in formaldehyde (4% wt/vol paraformaldehyde in 0.1 mol/L of sodium cacodylate buffer, pH 7.4) at room temperature (RT) and transported to the laboratory within 2 hours. The specimen is examined in a dissecting microscope. The fresh specimen is placed in optimal cutting temperature medium (Tissue-Tek O.C.T. compound; Sakura Finetek Europe BV, Zoeterwude, The Netherlands) on a piece of cork, snap frozen in isopenthane precooled with solid carbon dioxide at ⫺70°C, and stored in small vials at ⫺70°C until cryostat microtome sectioning. From the formaldehyde-immersed specimen, small parts, approximately 1 to 2 mm in diameter, of cortical tissue are collected and processed separately for electron microscopy. For further fixation, the container with renal tissue is placed on a shaking table at RT, and the total fixation time is recorded. Dehydration in ethanol-xylene and embedding in paraffin (reagent grade; melting point, 65°C) is carried out with a total processing time of 2.5 (minimum) to 4 to 6 hours (standard) in a programmable automatic tissue processor (VIP; Miles Scientific, Inc, Corona, CA). The final paraffin blocks are stored at RT.

Light Microscopy For LM, serial paraffin sections, 3 to 4 ␮m in thickness, are produced at 3 levels. Slides are deparaffinized with xylene and ethanol in a programmable stainer (DRS-601; Sakura Finetek USA, Inc, Torrance, CA). All biopsy specimens are stained with vanGieson, eosin, elastin, Jones silver (1 slide each), and periodic acid–Schiff, a total of 18 sections from each biopsy. Extra slides are kept for supplementary staining, if needed. We follow the World Health Organization classification of glomerular diseases2 and the Banff 97 upgraded working classification of renal allograft pathology.30

Reagents/Equipment for Immunostaining For protease digestion, we use Protease type XXIV (P8038; Sigma, St Louis, MO), 500 ␮g/mL, with calcium chloride, 500 ␮g/mL, in trihydroxymethylaminomethane hydrochloric acid–buffered saline (TBS), 0.5 mol/L, pH 7.8, at 37°C in a heated cabinet. IP staining is performed on a TechMate 500

676

MÖLNE, BREIMER, AND SVALANDER

(Dako A/S, Copenhagen, Denmark) computer-controlled stainer. All reagents necessary, except primary antibodies, are supplied in the Dako ChemMate EnVision Detection Kit (code no K5007). The detection system used, Dako EnVision HRP, is an avidin-free 2-step indirect method with secondary antibodies (goat-antirabbit and mouse immunoglobulins) and HRP conjugated to a dextran back bone.28 In the present work, IF and IP staining included the demonstration of immunoglobulin G (IgG), IgA, IgM, C1q, and C3c by using FITC-labeled and unlabeled primary antibodies supplied by Dako, respectively, as listed in Table 1.

IF Technique Optimum antibody dilutions were predetermined originally by means of titration and testing on series of sections from positive cases with glomerular disease. The highest dilution giving a clearly positive signal was used. Negative controls were obtained by omitting or replacing the primary antibodies. Negative tissue controls were baseline biopsy specimens from living-related kidney donors. Direct (1-layer) IF with fluorescein isothiocyanate (FITC)labeled antibodies is performed manually with slight modifications of our technique, described earlier.31 Briefly, series of cryostat microtome sections, 3 to 4 ␮m in thickness, are collected on clean glass slides and air dried at RT. After rehydration and washing (2 times at 5 minutes) in phosphatebuffered saline, pH 7.6, sections are incubated with FITClabeled antibodies at predetermined dilutions (Table 1) in a humid dark chamber for 30 minutes at RT. After washings in phosphate-buffered saline (3 times at 5 minutes) and aqua destillata (5 minutes) and drying, slides are mounted with Entellan neu (Merck, Darmstadt, Germany). Until viewing, slides are stored shortly in the refrigerator at ⫹4°C. Specimens are examined on a Nikon research microscope (Nikon Corporation, Tokyo, Japan) with a xenon 50-W lamp, with incident illumination and the appropriate filters for FITC-fluorescence. The intensity of IF staining is scored by a 4-level scale: 0 indicates no detectable staining; trace, small amount without homogenous pattern; 1⫹, distinctly positive; and 2⫹, strongly positive. Observations are recorded with a digital SPOT camera (Diagnostic Instruments Inc, Sterling Heights, MI), used for IF as well as IP and LM documentation. Digital micrographs captured in

Adobe Photoshop software (Adobe Systems Inc, San Jose, CA) are printed on an Epson Stylus Photo 870 ink jet printer (Epson Ltd, Hempstead, UK).

IP Technique The IP method used here is an indirect technique with unlabeled primary antibodies and the Dako EnVision HRP detection reagent in the second step. The procedure is standardized with a total processing time of 2.5 hours. The most important steps are as follows. (1) Consecutive series of paraffin sections are produced, preferably with a rotary microtome, at a 4-␮m constant thickness setting, floated on a 37°C water bath, and collected on serially numbered polylysin coated glass slides (Dako). Air-drying in a heated cabinet at 37°C for a minimum of 1 hour facilitates the adhesion of sections. (2) Sections are deparaffinized in xylene-ethanol at RT and rehydrated in TBS at 37°C immediately before protease treatment. (3) Proteolytic enzyme digestion is a critical step in the procedure. Freshly prepared Protease type XXIV, 500 ␮g/mL, in TBS, pH 7.8, at 37°C for 20 minutes works well. Digestion is terminated by repeated washing in TBS at RT. In the case of extended fixation for 24 hours or longer, digestion time usually is increased to 40 minutes. (4) Endogenous peroxidase activity is blocked by immersion in peroxidase-blocking solution (Dako S2023) for 5 minutes at RT. (5) Immunostaining is performed in a computer-assisted automatic TechMate 500 processor (Dako). Incubation time for primary antibodies is 30 minutes at RT. This step is terminated by repeated washings. (6) Slides are transferred to fresh hydrogen peroxide plus 3-3-diaminobenzidine tetrahydrochloride solution for 4 minutes. (7) Finally, slides are stained with Mayer’s hematoxylin and permanently mounted under cover slips. Negative controls are produced by omitting or replacing the primary antibodies. The intensity of IP staining is recorded in a 4-level scale, and results are documented as described for IF.

External Qualification Assessments Our immunohistochemical laboratory is affiliated with the UK National External Qualification Assessment Scheme for Immunocytochemistry at the Department of Histopathology, University College, London Medical School.32

Table 1. Antibodies Applied for IF and IP Staining IF

IP

Antigen Detected

Code No.*

Antibody (mg/L)

Dilution†

Code No.*

Antibody (mg/L)

Dilution‡

IgG IgA IgM Clq C3c

F0315 F0204 F0203 F0254 F0201

200 200 400 600§ 400

1:40 1:40 1:40 1:10 1:20

A0423 A0092 A0425 A0136 A0062

2,500 3,300 5,300 3,700§ 7,000

1:100,000 1:100,000 1:50,000 1:4,000 1:8,000

*Antibodies supplied by Dako A/S. †Phosphate-buffered saline, pH 7.8. ‡Antibody diluent, Dako code no. S 2022. §Total protein concentration.

IMMUNOPEROXIDASE STAINING OF RENAL BIOPSIES

Diagnostic Categories Examined In the present study, 81 renal biopsy specimens were examined. They were from our archive of approximately 800 native kidney biopsies assessed by means of LM and IF, processed during 1998 to 2004. All biopsy specimens were properly fixed and had a sufficient amount of renal tissue to allow for extended processing, but otherwise were randomly selected. This material included 70 cases of various glomerular diseases, ie, IgA nephropathy, n ⫽ 16; membranous nephropathy, n ⫽ 13; systemic lupus nephritis, n ⫽ 11; postinfectious glomerulonephritis, n ⫽ 11; anti–glomerular basement membrane (GBM) nephritis, n ⫽ 5; minimal change nephrosis, n ⫽ 4; focal and segmental glomerulosclerosis, n ⫽ 3; segmental necrosis with crescents (vasculitis), n ⫽ 2; mesangial proliferation, n ⫽ 2; membranoproliferative glomerulonephritis type I, n ⫽ 1; thin GBM disease, n ⫽ 1; and fibrillosis, n ⫽ 1. In addition, 11 cases of various other renal diseases were analyzed, including nephrosclerosis, n ⫽ 5; diabetes mellitus, n ⫽ 2; interstitial nephritis, n ⫽ 2; and thrombotic microangiopathy and amyloidosis, 1 case each. An extended series of paraffin sections were produced from each case for IP staining, as described. After IP staining, they were assessed separately, without knowing IF results. Scores for IF and IP were compared. IF and IP scores were divided into 2 groups in which scores of 0 and trace were recorded as negative and 1⫹ and 2⫹ were recorded as positive.

Statistical Analysis All recordings were assembled in a 2 ⫻ 2 contingency table of paired samples in which IF was regarded as the reference test (Fig 1). With the null hypothesis that there is no difference between IF and IP, the significance of differ-

Fig 1. For each staining, the score is recorded as positive reaction present (ⴙ) or absent (ⴚ), see text. Data are assembled in a 2 ⴛ 2 table for IF and IP. Because proportions are paired, ie, data refer to the same tissue studied by IF and IP, respectively, the comparison is based on frequencies of pairs with different outcomes.33 Presuming that IF is the test standard, IP sensitivity ⴝ a/(a ⴙ c), specificity ⴝ d/(b ⴙ d), concordance ⴝ a ⴙ d, and discordance ⴝ b ⴙ c. The test statistic is calculated with the continuity correction included and referred to the chi-square distribution for probability with 1 df (McNemar’s test), chisquare ⴝ (|b ⴚ c| ⴚ 1)2/(b ⴙ c).

677

ences was calculated as McNemar’s test using continuity correction for small samples.33

RESULTS

Tissue Treatment We have used our direct IF method for renal biopsy specimens in the routine for more than 30 years. Results are highly reproducible and continuously checked by negative and positive controls in our daily work with renal biopsies. Examples of IF findings documented in the renal biopsy specimens studied here are shown in Fig 2. With the IP technique, our results confirm that protease digestion is mandatory for the demonstration of immunoglobulins and complement factors in formaldehyde-fixed tissue (Fig 2). Without or with insufficient digestion, IP staining failed (not shown). By recording fixation time, we found a correlation between duration of tissue fixation and extent of proteolysis needed to get a positive staining. By slow rotatory agitation, 60 minutes of fixation at RT gave adequate LM preservation of renal core biopsy specimens. Without agitation, a minimum of 4 hours was needed. In our system, proteolysis for 20 minutes permitted a positive IP reaction in these tissues. Even after 4 to 18 hours of fixation, this treatment was effective. Fixation times of 24 hours or longer often required increased proteolysis, with a maximum of 40 minutes, after which tissue structure started to deteriorate (not shown). Because of the high number of coupled molecules, the Dako EnVision HRP is a highly efficient indicator system, permitting high dilution of the primary antibodies, ie, 1:4,000 to 1:100,000, with an incubation time of 30 minutes at RT. In comparison, dilutions of 1:10 to 1:40 are needed in our routine direct (1-layer) IF technique to give a satisfactory result (Table 1). The technical quality of IP staining was assessed in the light microscope. Specimens were accepted only when there were no stainable serum proteins in glomerular and peritubular capillaries. This was found to be a sign of adequate proteolysis and, indirectly, an indication of effective antigen retrieval. Conversely, if structural damage was present, eg, dissolution of cytoplasm and/or cell nuclei, indicating excessive digestion, the specimen was discarded. These problems occurred in less than 10% of IP-stained slides. They were corrected by increasing or

678

MÖLNE, BREIMER, AND SVALANDER

IMMUNOPEROXIDASE STAINING OF RENAL BIOPSIES

679

reducing digestion time, respectively. In addition, it was found that a constant section thickness was very important; 4 ␮m is optimal in the system used here. In 3 cases, single IP sections were damaged or incomplete; therefore, the corresponding incubations, n ⫽ 7, were not included in the evaluation, reducing the total number of IF/IP assessments to 398 (Table 2).

and always present as a diffuse global reaction in glomerular capillary walls (Fig 3D).

Morphological Characteristics The general staining pattern in the various types of glomerular diseases analyzed was very similar for the IF and IP techniques (Fig 2). However, by the IP technique, antigens were visualized in the light microscope and therefore could be correlated to tissue structures. Thus, in membranous nephropathy, membranous IgG deposits were clearly related to the capillary walls in the glomerulus (Fig 2, a2). Likewise, in IgA nephropathy, IgA deposits could be determined confidently as located in the mesangium (Fig 2, b2). Similarly, different patterns of C3c deposition could be discerned in postinfectious glomerulonephritis (Fig 2, c2) and membranoproliferative glomerulonephritis (Fig 2, d2). In addition, in cases of nephrotic syndrome with heavy proteinuria, plasma proteins often were detected by IP as resorption granules in the proximal tubular epithelium not always seen by IF (Fig 2). In several cases of minimal change nephrosis, we detected mesangial deposits of IgM by IP despite negative IF results (Fig 2, e2). In the present series, C1q was found in combination with IgM in 32 of 79 biopsy specimens (40%) and isolated in 13 of 79 biopsy specimens (16%) despite negative IF results (not shown). Results obtained by IF and IP are listed in Table 2. Tissues with a delayed start of fixation, exceeding 5 minutes, showed diffuse nonspecific staining of plasma proteins adhering to the tissue, not removed by standard protease-digestion (Fig 3). This phenomenon was most obvious with IgM

Statistical Evaluation Concordance between IF and IP staining was found in 160 and 122 of 398 observations (70.9%; Table 2). Using a 2 ⫻ 2 table of paired samples, IP staining showed a high specificity for IgG and IgA (0.89 and 0.94, respectively) and a lower value (0.8) for C3c. By McNemar’s test, stainings showed no statistically significant difference between IF and IP (P ⬎ 0.2). For IgM and C1q, sensitivity was high (0.98 and 1.0, respectively), but specificity was low (0.24 to 0.15), and there was a significant difference with positive staining by IP not obtained by IF (P ⬍ 0.001). For C3c, this difference was not significant (P ⬎ 0.2). Among discordant results (116 of 398 observations; 29%), 91 observations were positive by IP and negative by IF, and 25 observations (6.3%) were positive by IF and negative by IP. The majority, 74 of 91, of discordant stainings with positive IP showed IgM and/or C1q. In 9 observations, IP showed C3c, which was lacking in IF specimens; in 5 cases, deposits were located in the mesangium (Fig 2, e2), and in 4 biopsy specimens of membranous nephropathy, deposits were located in the capillary walls, whereas only a trace was seen in IF. C3c positive by IP alone occurred in a variety of glomerular diseases, some of which showed an increase in mesangial matrix to a variable degree; generally, only small amounts were found, which did not affect the diagnosis. The 9 discordant observations with positive IF results had focal and segmental deposits of C3c in 5 (vasculitis, fibrillosis, postinfectious, and 2 cases of IgA nephropathy) and 4 cases (anti–GBM disease and systemic lupus, 2 cases each), located diffusely in the capillary walls and detected by IF only (not shown).

4™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™ Fig 2. A panel of typical cases of glomerular diseases evaluated by (left) LM and corresponding findings by (central) IP and (right) IF. Row a, membranous glomerulonephritis stained for IgG; row b, IgA nephropathy stained for IgA; row c, postinfectious proliferative glomerulonephritis; row d, membranoproliferative glomerulonephritis, both stained for C3c; row e, minimal change nephrosis stained for IgM. Left column (1), periodic acid–Schiff– stained paraffin sections of formaldehyde-fixed renal biopsy specimens; central column (2), IP-stained sections from the same biopsy specimen; right column (3), IF on frozen sections from the same case. Note the similarity of the IH pattern between IP and IF in each case. The last row, e, shows a case of minimal change nephrosis in which IP shows faint mesangial deposits of IgM not found by IF. Note that by IP, the tissue structure is visible, permitting correlation between structure and IH reaction not evident by IF.

680

MÖLNE, BREIMER, AND SVALANDER Table 2. Renal Biopsy Specimens Stained for IgG, IgM, C1q, IgA, and C3c by IF and IP Techniques

IF/IP

a (⫹/⫹)

b (⫺/⫹)

c (⫹/⫺)

d (⫺/⫺)

No.

Sensitivity* a/(a ⫹ c)

Specificity† d/(b ⫹ d)

Chi-Square‡

P

IgG IgA IgM C1q C3c ⌺

23 24 40 26 47 160

5 3 29 45 9 91

9 6 1 0 9 25

42 48 9 8 15 122

79 81 79 79 80 398

0.72 0.8 0.98 1.0 0.84 0.86

0.89 0.94 0.24 0.15 0.63 0.57

0.64 0.44 24.3 43.02 0.05 36.42

⬎0.2 ⬎0.2 ⬍0.001 ⬍0.001 ⬎0.2 ⬍0.001

NOTE. Data recorded as paired proportions. C1q (n ⫽ 79), IgA (n ⫽ 81), and C3c (n ⫽ 80). *The proportion of positive results correctly identified by IP, presuming that IF is the test standard. †The proportion of negative results correctly identified as such. ‡The test statistic is calculated with the continuity correction included and referred to the chi-square distribution for probability with 1 df (McNemar’s test).

DISCUSSION

Antigen retrieval by proteolytic digestion is required for immunostaining of formaldehyde-fixed paraffin-embedded renal tissue, especially for the demonstration of immunoglobulin heavy and light chains and complement components in immunecomplex deposits.13,14 Although the exact mechanisms involved are not known and despite method disparity, this technique has been used successfully by several workers.15,34-37 For retrieval of cellsurface and nuclear antigens, high-temperature treatment with microwaves or pressure-cooking is preferred.38 Our routine fixative is formaldehyde (4%) made up from paraformaldehyde, which ideally gives a clean solution of hydrated formaldehyde (monomere) in sodium cacodylate buffer. By agitation, diffusion velocity is enhanced considerably, and for core biopsy specimens, immersion for 1 hour is sufficient. This secures a rapid and uniform fixation process, which we consider a prerequisite for controlled antigen retrieval. However, in several cases, we have been able to detect immune deposits in archival paraffin-embedded tissue originally fixed in formalin (unpublished results). The extent of digestion must be adjusted to the degree of formaldehyde fixation and section thickness, which should be kept constant. In our system, we found that Protease XXIV is superior to pronase and trypsin for reproducibility and safety. In accordance with Howie et al34 and Furness and Boyd,36 we found that successful removal of serum proteins normally present in vessels can be taken as an indication of a sufficient degree of protease digestion. Conversely, excessive digestion, indicated by a loss of tissue integrity, eg, by loss of cytoplasm and nuclear

structure, will spoil the preparation. The earlier peroxidase-labeled antibody and unlabeled peroxidase plus antiperoxidase immunobridge techniques have been replaced successively by more effective methods based on avidin-biotin binding,39 of which several variants are in use today.11 In avidin-biotin detection systems, the endogenous biotin activity present in many tissues and certainly in the renal tubular epithelium may give an unspecific background staining.40 In this report, the detection system contains secondary antibodies and HRP conjugated to a dextran backbone, omitting the biotin problem. The principle of this system has been detailed elsewhere.28 Another problem to keep in mind when evaluating IP-stained renal biopsy specimens is the importance of immediate fixation. In tissue with a delay of more than a few minutes before fixation, there is a diffuse positive reaction of plasma proteins that, in our experience, cannot be removed by the protease treatment. To our knowledge, there are only 3 published studies comparing IF and IP of renal biopsy specimens in such detail that statistical comparison with our results is possible (Table 3). In these reports, no statistical data were presented. However, calculated from the numbers given, their concordance data (range, 82% to 88%) differ from our findings (70.9%). Using a 2 ⫻ 2 table for paired samples, sensitivities in 2 of these studies were 0.8813 and 0.98,14 close to our findings of 0.86. In the third study, sensitivity is only 0.73.34 However, they used 2 different IP methods in parallel, namely peroxidase plus antiperoxidase for IgG, IgA, and IgM and an indirect method with HRP-labeled antibodies for C3c

IMMUNOPEROXIDASE STAINING OF RENAL BIOPSIES

681

Fig 3. Two renal biopsy specimens from a patient with nephrotic syndrome, stained with (top) periodic acid–Schiff and (bottom) IP for IgM. (A, C) One biopsy specimen was treated according to standard, whereas (B, D) the other had a delay of 30 minutes before formaldehyde fixation was started. The periodic acid–Schiff sections show (A) well-fixed tissue and (B) artifacts caused by delayed fixation. IP staining shows (C) a clear mesangial pattern of IgM in the immediately fixed tissue and (D) a diffuse global reaction in the delayed specimen.

studies by MacIver et al13 and Sinclair et al.14 In the study by Howie et al,34 and the present study (Table 2), the corresponding difference is significant (P ⬍ 0.001). However, this is not valid for

(C1q was not assessed). This may explain, in part, the discrepancy. In the overall results, there is no statistically significant difference between IF and IP in the

Table 3. Reported Studies Comparing IF and IP in the Assessment of Renal Biopsy Specimens Reference

IP

No.*

Concordance (%)

Sensitivity

Specificity

P (McNemar)†

MacIver et al,13 1979 Sinclair et al,14 1981 Howie et al,34 1990 Present work, 2004

PAP PAP PAP‡ EnV

119 147 260§ 398

82 83 88§ 70.9

0.88 0.98 0.73 0.86

0.75 0.55 0.95 0.57

⬎0.2 ⬎0.2 ⬍0.001 ⬍0.001

NOTE. Statistics calculated from published data, using a 2⫻2 table for paired proportions. Abbreviations: PAP, peroxidase and antiperoxidase; EnV, Envision HRP. *Total number of observations (IgG, IgA, IgM, C1q, and C3c). †Significance of difference in IF/IP calculated by McNemar’s test. ‡Indirect method with HRP-labeled secondary antibodies used for C3c. §Staining for C1q not performed.

682

the individual IP staining of IgG and IgA, which are equal to IF in both studies. This discrepancy depends on the proportion of pairs with different outcomes, ie, for Howie et al.34 C3c that often was undetected by IP and, in the present study, IgM and C1q that were detected more often by IP. The discordant findings were noted by the earlier investigators and regarded as without significance for the diagnosis. Because their techniques and scoring of findings were different from ours, these results are difficult to compare. We found stainings diverging both ways, ie, for the same antigen, IF and IP could give opposite positive or negative results, giving the impression that neither method is infallible. However, this discordance is not necessarily an indication that IP is an inferior method. More likely, it means that certain factors are removed during the processing and therefore not retrieved in the IF specimen. Consequently, IP on fixed tissue will detect antigens that may be lost or removed in the procedure in IF on unfixed tissue, especially during the washing steps. This is a very intriguing technical and diagnostic problem discussed by Brandtzaeg9 and others. However, one problematic diagnostic group, namely, the detection of linear deposits of IgG in the case of anti-GBM nephritis, is uncertain and/or defective with the IP technique, according to Howat et al.35 Their tissue handling includes washing fresh tissue in saline before fixation in formalin and digestion in trypsin, which is different from our method. In our material, we had 5 cases of anti-GBM nephritis, 2 of which showed linear deposition of IgG with IP, whereas the remaining 3 biopsy specimens were negative. The reason for this failure is not obvious at the moment. Although IF has retained its position as the gold standard, certain disadvantages are inherent in the original IF technology; namely, (1) the demand for fresh tissue, which requires 1 separate specimen, special handling, transportation, and treatment; (2) the need for special microscope equipment for fluorescence detection; (3) nonpermanence of the specimen because of fading and disintegration, which requires immediate documentation of observations and precludes permanent archiving; and (4) interpretation of findings that requires a certain level of skill and experience because of difficulties met with, eg, tissue autofluorescence and the need for pattern

MÖLNE, BREIMER, AND SVALANDER

recognition because nothing except positive reactions are visualized.15 In addition, from a clinical point of view, the need to take 1 separate biopsy core for IF in excess of material for LM, to be delivered fresh as soon as possible, is of major concern, especially when the laboratory is at some distance. Use of the transport medium of Michel et al19 may partly facilitate this procedure. However, additional needle passes increase the risk for complications caused by the biopsy procedure, ie, bleeding and tissue damage. IP techniques are advantageous compared with IF in several important respects; namely, (1) LM and IH can be performed and correlated on the same specimen, eliminating the need for initial special treatment, ie, cryotechnique; (2) end products are permanent, permitting archiving and later reviews of the slides15; and (3) later processing of stored paraffin-embedded tissue is possible. In addition, in single kidneys and other high-risk patients, one adequate biopsy only is sufficient for both LM and IH and even electron microscopy. We believe the standardized IP method presented here is progress toward the elimination of IF as the only IH method qualified for renal biopsies. Each step in the procedure, including the initial handling of the biopsy specimen, is of importance and must be controlled and standardized for reliable and reproducible IP staining. Our results indicate that, with one exception, glomerular deposits of immunoglobulin and complement of importance for the diagnosis in the assessment of our renal biopsy specimens were detected by IP. The exception is linear deposits of IgG that could not be shown in all cases of anti-GBM nephritis. REFERENCES 1. Pirani CL: Evaluation of kidney biopsy specimens, in Tisher CC, Brenner BM (eds): Renal Pathology (ed 2). Philadelphia, PA, Lippincott, 1994, pp 85-115 2. Churg J, Bernstein J, Glassock RJ (eds): Renal Disease, Classification and Atlas of Glomerular Diseases (ed 2). Tokyo, Japan, Igaku-Shoin, 1995 3. Jennette JC, Olson JL, Schwartz MM, Silva FG (eds): Heptinstall’s Pathology of the Kidney (ed 5). Philadelphia, PA, Lippincott-Raven, 1998 4. Neilson EG, Couser WG (eds): Immunologic Renal Disease. Philadelphia, PA, Lippincott-Raven, 1997 5. Coons AH, Creech HJ, Jones RN, Berliner E: The demonstration of pneumococcal antigen in tissues by the use of fluorescent antibody. J Immunol 45:159-170, 1942

IMMUNOPEROXIDASE STAINING OF RENAL BIOPSIES

683

6. Kashgarian M: Immunecomplex, in True LD (ed): Atlas of Diagnostic Immunohistopathology. Philadelphia, PA, Lippincott, 1990, pp 17.11-17.35 7. Heyderman E: Immunoperoxidase technique in histopathology: Application, methods and controls. J Clin Pathol 32:971-978, 1979 8. DeLellis R, Sternberger L, Mann R, Banks P, Nakane P: Immunoperoxidase techniques in diagnostic pathology. Am J Clin Pathol 71:483-488, 1979 9. Brandtzaeg P: Tissue preparation methods for immunohistochemistry, in Bullock GR, Petrusz P (eds): Techniques in Immunocytochemistry. Orlando, FL, Academic, 1982, pp 1-75 10. Turner D, Wilson D, Lake A, Heaton J, Leibowitz S, Cameron J: An evaluation of the immunoperoxidase technique in renal biopsy diagnosis. Clin Nephrol 11:13-17, 1979 11. Dodson A: Modern methods for diagnostic immunocytochemistry. Curr Diagn Pathol 8:113-122, 2002 12. Bishop P: An immunohistochemical vade mecum. Curr Diagn Pathol 8:123-127, 2002 13. MacIver A, Giddings J, Mepham B: Demonstration of extracellular immunoproteins in formalin-fixed renal biopsy specimens. Kidney Int 16:632-636, 1979 14. Sinclair R, Burns J, Dunnill M: Immunoperoxidase staining of formalin-fixed, paraffin-embedded human renal biopsies with a comparison of the peroxidase-antiperoxidase (PAP) and indirect methods. J Clin Pathol 34:859-865, 1981 15. MacIver M, Mepham B: Immunoperoxidase techniques in human renal biopsies. Histopathology 6:249-267, 1982 16. Griffiths M, Miller K: Peroxidase immunocytochemistry on formalin fixed, routinely processed paraffin-wax embedded renal biopsies: A method for demonstrating immune-complexes in glomerulonephritis. UK NEQAS Immunocytochem News, 1995/96, pp 2-5 17. Mepham B, Frater W, Mitchell B: The use of proteolytic enzymes to improve immunoglobulin staining by the PAP technique. Histochem J 11:345-357, 1979 18. Dische F: Renal Pathology (ed 2). Oxford, England, Oxford University Press, 1995, p 4 19. Michel B, Milner Y, David K: Preservation of tissuefixed immunoglobulins in skin biopsies of patients with lupus erythematosus and bullous diseases. A preliminary report. J Invest Dermatol 59:449-452, 1973 20. Sainte-Marie G: A paraffin embedding technique for studies employing immunofluorescence. J Histochem Cytochem 10:250-256, 1962 21. Heron I: A paraffin embedding method in kidney immunofluorescent studies. APMIS 78B:444-450, 1970 22. Dorsett B, Ioachim H: A method for the use of immunofluorescence on paraffin-embedded tissues. Am J Clin Pathol 69:66-72, 1978 23. Bolton W, Mesnard R: New technique of kidney tissue processing for immunofluorescence microscopy: Formol sucrose/gum sucrose/paraffin. Lab Invest 47:206-213, 1982 24. Svalander C, Eggertsen G, Olding L: Conditions for the immunohistochemical demonstration of complement fac-

tor C3 in formaldehyde fixed and paraffin embedded tissue. Histochemistry 84:81-85, 1986 25. Breimer M, Björck S, Svalander C, et al: Extracorporeal (“ex vivo”) connection of pig kidneys to humans. I. Clinical data and studies of platelet destruction. Xenotransplantation 3:328-339, 1996 26. Bengtsson A, Svalander C, Mölne J, Rydberg L, Breimer M: Extracorporeal (“ex vivo”) connection of pig kidneys to humans. III. Studies of plasma complement activation and complement deposition in the kidney tissue. Xenotransplantation 5:176-183, 1998 27. Rydberg L, Mölne J, Strokan V, Svalander C, Breimer M: Histo-blood group A antigen expression in pig kidneys. Scand J Urol Nephrol 35:54-62, 2001 28. Vyberg M, Nielsen S: Dextran polymer conjugate two-step visualisation system for immunohistochemistry. Appl Immunohistochem 6:6-30, 1998 29. Sabattini E, Bisgaard K, Ascani S, et al: The EnVision⫹ system: A new immunohistochemical method for diagnostics and research. Critical comparison with the APAAP, ChemMate, CSA, LABC and SABC techniques. J Clin Pathol 51:506-511, 1998 30. Racusen L, Solez K, Colvin R, et al: Antibodymediated rejection criteria—An addition to the Banff 97 classification of renal allograft rejection. Am J Transplant 3:708-714, 2003 31. Schmekel B, Svalander C, Bucht H, Westberg N: Mesangial IgA glomerulonephritis in adults. Acta Med Scand 210:363-372, 1981 32. UK National External Quality Assessment Service, Immunohistochemistry, Department of Histopathology, University College, London Medical School. Available at: www.ukneqas.org.uk. 33. Altman D: Practical Statistics for Medical Research. London, England, Chapman & Hall, 1991, pp 258-259 34. Howie A, Gregory J, Thompson R, Adkins M, Niblett A: Technical improvements in the immunoperoxidase study of renal biopsy specimens. J Clin Pathol 43:257-259, 1990 35. Howat A, Thomas C, Coward R: Immunoperoxidase for the demonstration of immune deposits in renal biopsies. Curr Diagn Pathol 6:125-129, 2000 36. Furness P, Boyd S: Electron microscopy and immunocytochemistry in the assessment of renal biopsy specimens: Actual and optimal practice. J Clin Pathol 49:233-237, 1996 37. Petrusz P, Ordronneau P, Finley J: Criteria of reliability for light microscopic immunocytochemical staining. Histochem J 12:333-348, 1980 38. Jasani B, Rhodes A: The role and mechanism of high-temperature antigen retrieval in diagnostic pathology. Curr Diagn Pathol 7:153-160, 2001 39. Hsu S, Raine L, Fanger H: Use of avidin-biotinperoxidase complex (ABC) in immunoperoxidase techniques: A comparison between ABC and unlabelled antibody (PAP) procedures. J Histochem Cytochem 29:577-580, 1981 40. Mount S, Cooper K: Beware of biotin: A source of false-positive immunohistochemistry. Curr Diagn Pathol 7:161-167, 2001