For optimal interpretation of BMT specimens, a combination of excellent cytomorphology, provision to use the full range of immunohistochemical stains, preservation of mRNA to carry out ISH analyses and the ability to extract DNA of a quality sufficient to carry out PCR‐based analyses are all crucial. Although it is generally accepted that resin embedding and semithin sections provide the best morphology, there have been questions on preservation of antigens and nucleic acids.2,6
Several papers deal with these issues and confirm adequate antigen and DNA preservation in resin‐embedded BMT specimens.7,8,9,10
However, RNA preservation and suitability for ISH analyses of mRNA is not clear. The main criticism of this procedure is the requirement of specialised technology and skill and the additional costs associated with it.
Optimal morphology without compromising preservation of antigens and nucleic acids
Our primary intention was to standardise a protocol for processing BMT specimens, which would result in cytomorphology of a quality comparable to that of resin‐embedded semithin sections and be suitable for a wide range of immunohistochemical staining. With this purpose, we embarked on AZF fixative, which is formalin based and contains the nuclear mordant zinc. We are able to cut 1‐μm‐thick sections providing excellent morphology (fig 1). To hasten the decalcification process, we use a short period of formic acid decalcification.
Figure 1(A–D) Haematoxylin and eosin (H&E) section of chronic myeloid leukaemia (A) ×40; (C) ×200 with foci of abnormal localisation of immature precursors; (B) ×100; (D) ×400. (E–H) Myeloproliferative (more ...)
EDTA, which is widely used for processing of BMT specimens, is slow and takes a minimum of 24 h (although application of an ultrasonic bath can help hasten this process).11
Although some laboratories report better success at DNA amplification on samples decalcified with EDTA, this is variable.12
On the other hand, acid decalcification with hydrochloric acid, nitric acid or formic acid is faster, but is believed to result in some destruction of antigens and tissue architecture. Short periods of acid decalcification, however, do not seem to have a deleterious effect on molecular diagnostics, including fluorescence in situ hybridisation and comparative genomic hybridisation studies.13,14
To obtain better morphology, fixatives like Bouin's, mercury‐containing solutions such as Zenker's fixative and B5 have been used. Some of these are, however, said to have a deleterious effect on preservation of antigens and nucleic acids, and mercurial fixatives pose problems related to disposal.15,16
Although the morphology is not comparable to that of semithin sections, formalin fixation provides superior antigen and DNA preservation as compared with picric acid or mercury‐based fixative.16,17,18
Furthermore, formalin‐fixed trephine sections provide adequate RNA preservation as seen in the few published studies on mRNA ISH.19,20
Thus, formalin fixation seems to be a good “compromise” fixative for processing of BMT specimens. The main criticism has been the suboptimal morphology as compared with resin‐embedded sections.
Zinc‐containing fixatives have been reported to be superior to formalin in terms of antigen preservation for immunohistochemistry and the recent study by Bonds et al4
clearly shows the superiority of AZF over other fixatives for preservation of morphology, antigens, RNA and DNA.4,21
The mechanism by which zinc interacts with proteins and nucleic acids is unclear and is currently part of our research programme in Imperial College. Formalin seems to be necessary for optimal morphology as our modified zinc‐based fixative (Z7) still results in some tissue shrinkage, but is compensated for by much improved protein and nucleic acid integrity.
Histochemical and immunohistochemical profiling and molecular diagnostics
In our hands, sections from AZF‐fixed paraffin‐wax‐embedded BMT specimens are optimal for Giemsa, silver and Perl's stains (fig 1). The process of decalcification reduces the amount of stainable iron in BMT specimens. Previous studies have reported that processing of BMT specimens with zinc–formalin fixative and formic acid decalcification results in an underestimation of marrow iron. It is only in undecalcified material with plastic sections that ring sideroblasts are easily identified.22,23
Because of the nature of BMT samples received at the Hammersmith Hospital (predominantly neoplastic) and treatment or management policies, we use immunohistochemical analysis on most cases (fig 2). Table 1 provides the details of the antibodies we use on BMT specimens. In suspected cases of myeloproliferative disorders and myelodysplastic syndromes, we routinely carry out CD34, CD117, ret40f, myeloperoxidase and CD42b/CD61. This helps us quantify the precursor cell population, identify foci of abnormal localisation of immature precursor cells and identify subtle dysplastic features in the various lineages, including identification of micromegakaryocytes.24,25,26
In addition, cases of systemic mastocytosis are immunostained for mast cell tryptase, CD25 and CD2.27,28
Cases of suspected acute leukaemia are investigated on a panel of antibodies that includes CD34, TdT, HLA‐DR, CD99, myeloperoxidase, CD117, CD68 (KP‐1 and PG‐M1), ret40f, CD61/CD42b, CD20, CD79a, CD3 and CD10.6,29,30,31,32,33,34
As well as helping in the proper classification of these cases, documentation of the leukaemia immunophenotype is of utmost importance in evaluating follow‐up BMT specimens for response to treatment, residual disease and early relapse.
Figure 2(A–F) Hairy‐cell leukaemia, haematoxylin and eosin (H&E) staining (A) ×40; (B) ×200; (C) TRAP, ×200; (D) CD25, ×200; (E) DBA44, ×200; (F) CD20, ×200. (G–K) (more ...)
On BMT specimens taken for staging histologically documented B cell non‐Hodgkin's lymphoma, we use a limited panel for immunohistochemistry—CD20, CD3 and CD79a.35
In cases of patients with B cell non‐Hodgkin's lymphoma who have received anti‐CD20 treatment, immunostaining with CD20 is unreliable and may even be negative in residual B cell lymphoma cells. In such cases, CD79a is valuable in identifying residual lymphoma. For staging BMT specimens in patients with anaplastic large‐cell lymphomas, the panel includes CD20, CD3, CD30, ALK, epithelial membrane antigen and Pax‐5.36
The panel for staging BMT specimens of other node‐based peripheral T cell lymphomas includes CD3, CD20, CD4, CD8, CD30, CD10 and CD21.6,37
In BMT specimens taken for staging Hodgkin's lymphoma and in those where involvement is suspected on the H&E stain, the panel includes CD20, CD3, CD15 and CD30. In difficult cases, an extended panel is required.
BMT specimens taken from patients with a clinical diagnosis of B cell chronic lymphoproliferative disorders, depending on the likely diagnosis, are evaluated for expression of CD20, CD3, CD79a, CD5, CD10, CD21, CD23, CD25, CD43, CD38, CD138, cyclin D1, bcl‐2, bcl‐6, Pax‐5, DBA44, TRAP, and immunoglobulin heavy and light chains.6,38,39,40,41,42,43,44
In patients with suspected T cell lymphoproliferative disorders, depending on the likely diagnosis, BMT specimens are evaluated for expression of CD2, CD3, CD5, CD7, CD4, CD8, CD20, CD56, CD57, cytotoxic molecules, CD30 and TCRβ.6,45
In patients with suspected clonal plasma cell proliferations, BMT specimens are evaluated for expression of CD20, CD79a, CD3, CD138, vs38C, CD56, cyclin D1, epithelial membrane antigen and immunoglobulin light chains and, in some cases, heavy chains.6
Documenting the phenotype of myeloma cells in each case, in our experience, has been extremely valuable in evaluating follow‐up samples for residual disease. The immunohistochemistry panel in BMT specimens evaluated for metastatic disease and histiocytic proliferations would depend on histological evaluation of the H&E section.
For the purposes of research, we have carried out double immunostains with antibodies to ret40f, CD61, Ki‐67 and Mcm‐2 on a good number of BMT specimens with very good results (fig 3).46
Figure 3Myeloproliferative disorder; Mcm‐2 in nucleus (red) and CD61 in cytoplasm of megakaryocytes (blue; ×400). The figure has been contributed by Dr I Lampert and is part of his ongoing research.
Over the past 6 months, we have been carrying out ISH analysis of mRNA for κ and λ light chains and of EBV‐encoded small RNA‐1 for EBV. The results have been good and mRNA preservation has been more than satisfactory (fig 4). Although there have been many studies investigating clonality in lymphoid or plasma cell proliferations with mRNA‐ISH for immunoglobulin light chains, only a few of the published studies have been on BMT specimens.47,48,49
Generally, immunostaining for light chains is adequate. In BMT samples, however, in cases which have abundant background staining owing to adsorbed immunoglobulin in the background tissues, ISH results are often cleaner and much easier to interpret. Few studies relating to ISH for EBV on BMTs exist. These can be of great value in identifying bone marrow involvement by EBV‐associated lymphoid malignancies such as T cell or natural killer cell lymphomas and Hodgkin's lymphoma.50
Although cells expressing CD30 or CD3 or other tumour cell‐marking antigens would not be restricted to the tumour cell population in the marrow sample, EBV would almost be restricted to tumour cells. ISH for EBV‐encoded small RNA‐1 is more useful than immunostaining for EBV‐LMP‐1 as it is more sensitive and is expressed across all types of EBV latency.51
The strategy would be very valuable in evaluating bone marrow samples from patients with HIV in whom 100% of cases of Hodgkin's lymphoma are EBV associated and the bone marrow is affected in a large proportion of patients. In our experience and that of others, bone marrow may often be the initial site of disease presentation.52
Figure 4(A,B) mRNA in situ hybridisation (ISH) with Epstein–Barr virus‐encoded small RNA‐1 probe in a case of Hodgkin's lymphoma focally affecting the bone marrow (A) ×100; (D) ×400. (C,D) mRNA‐ISH (more ...)
More recently, we extracted DNA from BMT specimens processed by our protocol for the purposes of PCR amplification. We carried out PCR analysis for immunoglobulin gene heavy‐chain rearrangement on a few samples (fig 5). We need to study cases for a wider array of PCR‐based molecular diagnostics and analyse a large number of samples. Although many investigators find formalin‐fixed, EDTA‐decalcified paraffin‐wax‐embedded BMT specimens to be adequate for molecular diagnostics, the preservation of DNA has been previously reported to be superior with zinc‐based fixatives compared with formalin.53,54
Figure 5PCR with DNA extracted from bone marrow trephine (BMT) specimens. Lanes 1 and 4, a BMT specimen with no evidence of neoplastic disease. Lanes 2 and 5, a BMT specimen heavily affected by chronic lymphocytic leukaemia. Lanes 3 and 6, water (more ...)
Laboratory integration and turnaround time
We aimed at making a protocol that would be quick, so as not to hamper the turnaround time and at the same time include procedures that would generally be similar to other tissue processing protocols in the laboratory. The H&E‐stained section with Giemsa, reticulin and Perl's and the initial panel of immunostains is ready in 72 h of receipt of BMT samples. The H&E‐stained section and the sections stained with histochemical stains are available in 48 h, if there is a need for urgent or provisional report (table 2).
Table 2Turnaround time for bone marrow trephine biopsy specimens including first panel of immunohistochemistry
It should be emphasised that thin sectioning, preferably by an experienced histotechnologist, is a key variable in ensuring excellent slides from which a maximal amount of diagnostic information can be extracted.