www.thelancet.com/hiv Vol 11 April 2024

我們懷著極大的興趣閱讀了 Denton Callander 及其同事關於愛滋病毒治療作為預防在澳洲同性戀、雙性戀和其他男男性行為者 (GBM) 中預防愛滋病毒發生率人群層級之有效性的文章。結論是，2010 年至2019 年間，GBM 族群中病毒抑制盛行率每增加1%，HIV 發生率就會下降5%。特別值得注意的是，分層分析結果分為暴露前預防 (PrEP) 推出之前 (2010-16) 和之後 (2016-19)。在 PrEP 前階段，病毒抑制率每增加 1%，HIV 發生率就會降低 2%。在 PrEP 期間，這種效應增加了驚人的十倍：病毒抑制率每增加 1%，HIV 發生率就會下降 20%。

        我們發現作者的建議「雖然治療作為預防和 PrEP 都是重要的公共衛生策略，但它們的真正潛力是結合起來釋放的」，這一建議並不令人滿意。考慮到堅持使用 PrEP 的使用者無論其性伴侶的病毒載量如何，都可以免受 HIV 感染，因此人們預計在引入 PrEP 後，治療作為預防對 HIV 發生率的影響將會降低。此外，效應量的增加顯得異常大，顯示分析中可能還有其他因素在起作用，而沒有考慮到。

        首先，我們懷疑 PrEP 造成的殘餘干擾因素。儘管分析針對 PrEP 使用進行了調整，但 PrEP 使用似乎被建模為個人層面變數。這種方法糾正了個人層面的 PrEP 有效性，但沒有正確考慮人群層面的 PrEP 有效性，人群層級的 PrEP 有效性可能遵循 S 形曲線，作為 PrEP 使用流行率的函數。因此，PrEP 期間效應大小的增加可能至少部分歸因於 PrEP 的人群層面有效性，而不是單獨的治療作為預防的人群層面有效性。

        其次，我們質疑，在病毒抑制盛行率最初很高且隨著時間的推移進一步增加有限的情況下，將愛滋病毒發生率的每年變化與病毒抑制盛行率的每年變化進行回歸是否合適。在這項研究中，到2016 年，盛行率已達85.91%，到2019 年則增至88.31%。可以想像，作者將愛滋病毒發生率下降每年的變化不大之部分原因，歸因於研究族群中病毒抑制率已經很高的情況。

        總之，我們認為2016 年至2019 年間，病毒抑制盛行率每增加1%，愛滋病毒發生率就會下降20%，這一估計過高，很可能是由於與人群層面PrEP 有效性相關的殘留干擾因素，也可能是由於將愛滋病毒發生率持續下降錯誤地歸因於病毒抑制盛行率每年有限的增長。我們讚揚作者對這一重要主題的分析，但敦促他們考慮 2016 年後效應量顯著增加的替代解釋。

Callander D、McManus H、Gray RT 等人。 HIV 治療預防及其對澳洲順性別同性戀、雙性戀和其他男男性行為者中 HIV 發生率的影響：一項為期 10 年的縱向世代研究。刺胳針愛滋病毒 2023； 10：e385–93。

MFSvdL 獲得了吉利德 (Gilead) 和葛蘭素史克 (GSK) 的資助以及默克 (Merck) 和 Novosanis 的顧問委員會費用，所有這些都支付給他的機構。 ESW 和 MCV 聲明不存在競爭利益。 ESW 和 MCV 對本通訊做出了同等貢獻。

*Eline S Wijstma、Manon C Vanbellinghen、Maarten F Schim van der Loeff ewijstma@ggd.amsterdam.nl 阿姆斯特丹公共衛生服務部傳染病部，1018 WT 阿姆斯特丹，荷蘭 (ESW, MFSvdL)；阿姆斯特丹， UMC 地點 ，阿姆斯特丹大學傳染病系，荷蘭阿姆斯特丹（MCV、MFSvdL）；阿姆斯特丹免疫學和傳染病研究所，荷蘭阿姆斯特丹（MCV、MFSvdL）；阿姆斯特丹， UMC 地點，阿姆斯特丹大學全球健康系、阿姆斯特丹全球健康與發展研究所，荷蘭阿姆斯特丹 (MCV)

作者回復

         我們歡迎 Eline S Wijstma 及其同事參與我們的愛滋病毒治療作為預防之研究，並感謝他們針對結果提出了一些有趣的問題。

        我們的結論是，提高人口層面的治療作為預防的覆蓋率與愛滋病毒發生率降低相關，並且在引入暴露前預防 (PrEP) 後，這種相關性得到加強。對此，Wijstma 及其同事提出，在我們的觀察期之後期，病毒抑製程度是如此之高，評估愛滋病毒發生率的變化是不合適的。

         我們承認該地區同性戀、雙性戀和其他男男性行為者 (GBM) 的治療採用率和病毒抑制率很高，估計 2019 年人群層面的病毒抑制率為 88·31%，由2015年的83. 41%提升上來。儘管從絕對值來看，近 5 個百分點的差異似乎並不是很大，但這一變化反映了愛滋病毒照護級聯的許多關鍵組成部分的多項改進。這些改進很可能在乍看之下以並不明顯的方式提高了以治療為預防上的人群有效性。

         根據我們文章中的圖3，我們發現GBM 中新診斷的個案開始愛滋病毒治療的人數在後期大幅增加，增加了36.30%（2016 年每100 人年60.82 人增加到2019 年每100 人年82.90 人）。在其他地方，我們記錄了該族群中從診斷到治療的中位數時間的變化，2016 年至2019 年間減少了20%， 而在文章中，我們發現持續病毒抑制的發生率在2016 年至2019 年期間增加了112·46% (由2016年每100人年158·49到2019年每100人年336·73)。這些措施顯示，與實施 PrEP 之前相比，較晚期擁有可檢測到病毒載量的性伴侶的機會要低得多。此外，如附錄中所強調的，病毒量非常高（即每毫升超過10,000 個RNA 拷貝）的未確診男性比例從2016 年到2019 年略有下降（83.12% 至76.74%），這也對傳播的可能性有影響。總之，以治療為預防的許多改進顯示，即使在以後的幾年裡，其影響也會不斷增加。此外，如附錄中所強調的，病毒量非常高（即每毫升超過10 000 個RNA 拷貝）的未確診男性比例從2016 年到2019 年略有下降（83.12% 至76.74%），這也意涵著對傳播可能性的影響。總而言之，以治療為預防的諸多改進顯示，即使在以後的幾年裡，其影響力也不斷增加。

         Wijstma 和同事也指出，儘管 PrEP 包含在我們的計算中，但它可能會對結果產生干擾影響。儘管有可能，但我們注意到，到研究期結束時，只有 36.38% 的 GBM 正在使用 PrEP，這意味著大多數人仍然意識到增加病毒抑制和減少病毒血症的好處。然而，我們承認，區分愛滋病毒治療作為預防和 PrEP 的獨特貢獻是很困難的。事實上，隨著 PrEP 的引入，許多（通常無法測量的）社會、行為和生物因素發生了變化，這無疑在這個故事中發揮了一些未被認識到的作用。

        最終，我們並不否認要分離多面向的公共衛生策略對現實世界的影響是複雜的，但我們堅持我們最初分析得出的主要結論：為了在人口層面上減少愛滋病毒，有必要在治療即預防和 PrEP這兩個方面進行投資。

RG 是獲得吉利德科學公司非財務和研究支持的計畫的共同研究員。 MS 獲得了吉利德科學公司和艾伯維公司的資助，用於由研究者發起的與本研究無關的研究。 DC 和 HM 聲明不存在競爭利益。

*Denton Callander、Mark Stoové、Hamish McManus、Rebecca Guy，代表 TAIPAN 調查員 d.callander@unsw.edu.au

新南威爾斯大學柯比研究所，悉尼，新南威爾斯 2050，澳洲（DC、HM、RG）；澳洲維多利亞州墨爾本伯內特研究所 (MS)

Treatment-as prevention and HIV incidence in men who have sex with men

www.thelancet.com/hiv Vol 11 April 2024

It is with great interest that we read Denton Callander and colleagues’ Article on the population-level effectiveness of HIV treatment-as prevention on HIV incidence among gay, bisexual, and other men who have sex with men (GBM) in Australia. The authors conclude that, between 2010 and 2019, a 1% increase in the population prevalence of viral suppression was associated with a 5% decrease in HIV incidence among GBM. Particularly noteworthy were the results from the analysis stratified into the period before (2010–16) and after (2016–19) pre-exposure prophylaxis (PrEP) became available. In the pre-PrEP period, a 1% increase in prevalence of viral suppression was associated with a 2% decrease in HIV incidence. In the PrEP period, this effect increased by a staggering ten-fold: for each 1% increase in prevalence of viral suppression, HIV incidence decreased by 20%.

We find the authors’ suggestion that “although treatment-as-prevention and PrEP are both important public health strategies, their true potential is unlocked in combination” unsatisfying. One would expect the impact of treatment-as-prevention on HIV incidence to decrease after the introduction of PrEP, considering that adherent PrEP users are protected against HIV regardless of their sex partner’s viral load. Furthermore, the increase in effect size appears unusually large, indicating there might be other factors at play not accounted for in the analysis.

First, we suspect residual confounding by PrEP. Although analyses were adjusted for PrEP use, it appears that PrEP use was modelled as an individual-level variable. This approach corrects for individual-level PrEP effectiveness, but does not properly account for population-level PrEP effectiveness, which likely follows an S-shaped curve as a function of PrEP use prevalence. As such, the increased effect size in the PrEP period might, at least in part, be attributable to the population level effectiveness of PrEP rather than the population-level effectiveness of treatment-as-prevention alone.

Second, we question whether regressing per-year change in HIV incidence against per-year change in prevalence of viral suppression is suitable in a setting where prevalence of viral suppression is initially high and further increase over time is limited. In this study, prevalence had reached 85·91% by 2016, and increased to 88·31% by 2019. It is conceivable that the authors attributed part of the decline in HIV incidence, driven by the already high prevalence of viral suppression in the study population, to a modest per-year change.

In summary, we consider the reported 20% decrease in HIV incidence for each 1% increase in prevalence of viral suppression between 2016 and 2019 an over estimation, most likely due to residual confounding related to population-level PrEP effectiveness, and possibly due to the misattribution of ongoing decline in HIV incidence to a limited per-year increase in prevalence of viral suppression. We commend the authors for their analysis on this vital topic, but urge them to consider alternative explanations for the stark increase in effect size after 2016.

Callander D, McManus H, Gray RT, et al. HIV treatment-as-prevention and its effect on incidence of HIV among cisgender gay, bisexual, and other men who have sex with men in Australia: a 10-year longitudinal cohort study. Lancet HIV 2023; 10: e385–93.

MFSvdL received grants from Gilead and GSK and advisory board fees from Merck and Novosanis, all paid to his institution. ESW and MCV declare no competing interests. ESW and MCV contributed equally to this Correspondence.

*Eline S Wijstma, Manon C Vanbellinghen, Maarten F Schim van der Loeff ewijstma@ggd.amsterdam.nl Department of Infectious Diseases, Public Health Service of Amsterdam, 1018 WT Amsterdam, Netherlands (ESW, MFSvdL); Amsterdam UMC location University of Amsterdam, Department of Infectious Diseases, Amsterdam, Netherlands (MCV, MFSvdL); Amsterdam Institute for Immunology and Infectious Diseases, Amsterdam, Netherlands (MCV, MFSvdL); Amsterdam UMC location University of Amsterdam, Department of Global Health, Amsterdam Institute for Global Health and Development, Amsterdam, Netherlands (MCV)

 Authors’ reply

We welcome the engagement with our study of HIV treatment-as-prevention by Eline S Wijstma and colleagues, and thank them for raising some interesting questions about the results.

We concluded that increasing population-level coverage of treatment-as-prevention was correlated with reduced HIV incidence and that this correlation was strengthened after the introduction of pre-exposure prophylaxis (PrEP). In response, Wijstma and colleagues suggest that it was inappropriate to assess changes to HIV incidence during the latter part of our observation period when levels of viral suppression were so high.

We acknowledge high levels of treatment uptake and viral sup pression among gay, bisexual, and other men who have sex with men (GBM) in this region, with an estimated population-level viral suppression of 88·31% in 2019, up from 83·41% in 2015. Although a difference of nearly 5 percentage points does not appear very large in absolute terms, this change reflects multiple improvements to many key components of the HIV cascade of care. Such improvements are likely to have enhanced the population effectiveness of treatment-as-prevention in ways not readily apparent at first glance.

Per figure 3 in our Article,  we found HIV treatment initiation among newly diagnosed GBM increased substantially—by 36·30%— in this later period (60·82 per 100 person-years in 2016 to 82·90 per 100 person-years in 2019). Elsewhere, we have documented changes to the median time from diagnosis to treatment among this population, which decreased by 20% between 2016 and 2019, whereas in the Article we found the incidence of sustained viral suppression increased by 112·46% during this period (158·49 per 100 person-years in 2016 to 336·73 per 100 person-years in 2019). These measures indicate that the chance of having a sexual partner with a detectable viral load was much lower in later years compared with before the implementation of PrEP. Further, as highlighted in the appendix, the proportion of undiagnosed men with very high viral loads (ie, exceeding 10 000 RNA copies per mL) decreased slightly from 2016 to 2019 (83·12% to 76·74%), which also has implications for likelihood of transmission. Taken together, these many improvements to treatment-as-prevention suggest its increasing impact even in later years.

Wijstma and colleagues also note that despite its inclusion in our calculations, PrEP might have had a confounding effect on the results. Although possible, we note that by the end of the study period only 36·38% of GBM were using PrEP, meaning the majority would still have been realising the benefits of increasing viral suppression and decreasing viraemia. We acknowledge, however, that teasing apart the distinct contributions of HIV treatment-as-prevention and PrEP is difficult. Indeed, many (often unmeasurable) social, behavioural, and biological factors changed with the introduction of PrEP, which undoubtedly play some unrecognised part in this story.

Ultimately, we do not deny it is complex to isolate the real-world effects of multifaceted public health strategies, but we stand by the primary conclusion derived from our original analysis: to reduce HIV at a population level, it is necessary to invest in both treatment-as-prevention and PrEP.

RG is a co-investigator on projects that have received non-financial and research support from Gilead Sciences. MS received funding from Gilead Sciences and AbbVie for investigator-initiated research unrelated to this study. DC and HM declare no competing interests.

*Denton Callander, Mark Stoové, Hamish McManus, Rebecca Guy, on behalf of the TAIPAN investigators d.callander@unsw.edu.au

The Kirby Institute, University of New South Wales, Sydney, NSW 2050, Australia (DC, HM, RG); The Burnet Institute, Melbourne, VIC, Australia (MS)