| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Letters |
1
Sanofi Diagnostics Pasteur, ZI du Léopha, rue d'Italie, 69780 Mions, France,
2
CNRS UMR 9921, Faculté de Pharmacie, 34060 Montpellier Cedex 1, France
a author for correspondence.
To the Editor:
In light of the recent article by Spencer et al. (1) as well as the editorial by Spencer (2), we feel obliged to clarify a few points concerning the performance of our thyroglobulin (Tg) immunoassay and make some more general comments.
In Fig. 1 of ref. 1, the legend reads "Serum Tg values measured by six different Tg methods." The six assays tested are clearly indicated in the legend but only four of them appear in the Figure. The one from our group, Tg IRMA Pasteur, is missing.
Surprisingly, the "relative sensitivities of the five different Tg methods as judged by the response in signal cpm," shown in Fig. 2 of ref. 1, are expressed as absolute signals given by the bound (B) antibody tracer (background subtracted). When antibodies are labeled with radioisotopes that have a relatively short half-life, the results should preferably be expressed as the bound fraction (B/B0) to take into account radioactive decay. In our case (the Sanofi Diagnostics Pasteur test has a 125I label), the difference between the two modes of representation is particularly striking because the assay was performed approximately at the expiration date of the tracer. All evaluations of our Tg immunoassay, without fail, underlined the very high analytical sensitivity of our assay due to the use of a selected combination of antibodies.
Furthermore, Mariotti et al. (3) were cited in ref. 1 as having defined the true minimum detectable concentration of Tg at 1.5 µg/L with the Tg IRMA Pasteur assay. We demonstrated, however, that a cutoff of 1 µg/L gives the best results in terms of clinical accuracy (4). This value is significantly higher than the minimum detectable concentration defined at the confidence limit of 95% for the calibration zero (i.e., mean + 2 SD) of 0.2 µg/L. Had this recommended cutoff been taken into account, some of the false Tg-negative patients would probably have been found to be Tg positive in Mariotti's study (3).
Finally, considering 131I scintigraphy as a possible "gold standard," we think that this procedure may not be completely specific (see Dadparvar et al. (5)). The only way to definitely distinguish a false Tg-negative from a true Tg-negative patient is by clinical follow-up. This data is not given in Mariotti's article.
In her editorial (2), Spencer cites Mariotti's argument (3) that our IRMA (4), which is based on epitopic selection, offers no clinical advantages. She goes on to report Mariotti's findings that serum Tg concentrations assayed by the Tg IRMA Pasteur method were inappropriately undetectable in patients with metastatic thyroid cancer when Tg was maximally stimulated by a high thyrotropin (TSH) concentration. Although the withdrawal of L-thyroxine (L-T4) therapy generally leads to an increase in the TSH concentration, Mariotti et al. do not indicate the TSH concentrations observed.
Regarding Tg autoantibody (TgAb) interference in Tg measurements, Spencer et al. report in their article (1) that Ruf et al. (6) observed a difference between TgAb specificity in patients with carcinoma and those with autoimmune thyroid disease. We would also like to mention the recent publication by Caturegli and Mariotti (7), who showed that in thyroid carcinoma, TgAb do not recognize different clusters from those recognized in autoimmune diseases (Hashimoto thyroiditis and Graves disease), as first suggested by Piechaczyk et al. (8). The same remark applies for the editorial (2).
In her editorial (2), Spencer questions the significance of antigen recovery studies involving purified Tg added to TgAb-positive sera. We do agree that there is no definitive proof as to the identity between endogenous (serum) Tg and the thyroid extract. Nevertheless, we clearly demonstrated that the use of a combination of selected monoclonal antibodies (directed against antigenic regions not recognized by TgAb) leads to a significant reduction in autoantibody interference as compared with conventional immunoassays (6)(9).
The impact of the improvements, obtained by epitopic dissection, on diagnostic performances will be assessed in the near future by exploring Tg in larger populations of TgAb-positive and -negative patients.
References
General Clin. Research Center 6602, Dept. of Medicine, Univ. of Southern California, 2025 Zonal Ave., Los Angeles, CA 90033
a author for correspondence.
To the Editor:
The letter of Calzolari et al. provides a further comment on the performance of serum Tg assays, specifically the Pasteur Tg IRMA. The authors quite correctly point out that the legend of the Fig. 1 shown in our recent article (1) was in error when it stated that six different Tg methods were tested, when in fact only four are shown. This data originated from our laboratory and did not include the Pasteur method because this method was not available in the US. In contrast, the recombinant human (rh) TSH-stimulated Tg response data shown in Fig. 2 were collaboratively generated in different laboratories in the US and Europe, one of which used the Pasteur Tg IRMA.
Calzolari et al. question the use of the signal–background, as compared with a signal/background, parameter for judging immunometric assay (IMA) sensitivity from the rhTSH response data shown in Fig. 2 (1). Although this is a valid criticism, recalculation of the data as the ratio only changes the sensitivity ranking of the Pasteur method from fifth to fourth with respect to both patients. Further, their comment regarding a concern for using an isotopic tracer close to its expiration date only emphasizes the advantages of nonisotopic signals.
Minimum detection limits and clinical cutoffs cannot be compared in absolute terms (µg/L) unless Tg assays are standardized on the new International Reference Preparation (CRM 457) (2). Calzolari et al. discuss minimum detection limits based on analytical sensitivity (95% confidence limits of the signal for the zero calibrator). Analytical detection limit is a clinically meaningless parameter when compared with functional sensitivity based on low range interassay precision—a concept now firmly established for serum TSH measurements (3). It is especially important for a Tg assay that functional sensitivity be used to establish the lower reporting limit since the typical clinical interval for using Tg to monitor patients with thyroid cancer is 6 to 12 months.
Calzolari et al. appropriately question the use of 131I scintigraphy as the "gold standard" for judging the clinical sensitivity and specificity of serum Tg measurements. Disparities between imaging and serum Tg do not usually reflect differences in TSH secretory status rather than differences in scan dose (4)(5) and the sensitivity of the Tg assay method (1). Indeed, studies now suggest that negative 131I uptakes in patients with detectable serum Tg usually represent falsely negative scans since posttreatment scans will often reveal disease in many patients (4)(5). Current concepts dictate that a detectable serum Tg is expected whenever a patient has unequivocal metastatic thyroid cancer, even when serum TSH is suppressed (6). The Marriotti study of the Pasteur Tg IRMA (7) included only TgAb-positive patients in the high TSH state before imaging. The failure to detect serum Tg by IRMA in such a patient is more likely to be due to TgAb interference causing underestimation of the serum Tg measurement than a submaximal TSH increase.
The epitope selection approach used to develop the Pasteur IRMA is an attractive concept. The question of whether the Tg epitope fingerprint typical of thyroid cancer is the same as that of autoimmune thyroid disease is not as important as the question of the clinical validity of serum Tg measurements made in TgAb-positive patients. Calzolari et al. conclude their remarks by claiming that their epitope selection approach leads to a significant reduction in autoantibody interference as compared with "conventional" IMAs. Tg recovery cannot support this claim in view of the gross discordance between the RIA and IMA serum Tg results for TgAb-positive patients with unequivocal evidence of persistent disease (33.1, range 1.2–92 vs <0.3, range <0.3–1.1 µg/L, RIA vs IMA respectively (1)) despite "appropriate" (>80%) recoveries with both the RIA and IMA methods.
Clearly the clinical reliability of serum Tg measurements made in TgAb-positive sera by IMA methodologies remains a major concern, whether or not an epitope selection approach is used. It behooves manufacturers to show that their methods provide serum Tg values in TgAb-positive patients that are concordant with clinical status and prompt an appropriate clinical response. Patients with persistent TgAb detected on long-term follow-up usually have persistent disease and have a higher risk of recurrences (8)(9). For such patients, the use of a Tg IMA method prone to TgAb interference causing underestimation could potentially have serious medicolegal consequences, as an inappropriately low serum Tg result may lead to a delay in the diagnosis and treatment of recurrent or metastatic disease. Because recoveries cannot be used to validate the reliability of serum Tg measurements made in TgAb-positive sera, we should question the current practice whereby laboratories continue to report serum Tg values for such patients.
References
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  F. Pacini, E. Molinaro, M. G. Castagna, L. Agate, R. Elisei, C. Ceccarelli, F. Lippi, D. Taddei, L. Grasso, and A. Pinchera Recombinant Human Thyrotropin-Stimulated Serum Thyroglobulin Combined with Neck Ultrasonography Has the Highest Sensitivity in Monitoring Differentiated Thyroid Carcinoma J. Clin. Endocrinol. Metab., August 1, 2003; 88(8): 3668 - 3673. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
