With the introduction of targeted agents into the treatment paradigm of chronic lymphocytic leukemia (CLL), outcomes for patients have markedly improved. With prolonged overall survival in the predominantly elderly population of CLL patients, second primary malignancies (SPM) become more relevant as competing causes of death. Prior analyses have consistently shown increased rates of SPM in patients with CLL in different treatment contexts when compared to the general population.^1-4 With scarce data comparing treated and untreated patients with CLL, it remains unclear whether this increase in SPM can be attributed to CLL-inherent immunodeficiency, shared environmental or genetic risk factors of CLL and SPM or cancerogenic effects of antineoplastic treatments.

To specifically assess the effect of CLL-directed treatment on SPM incidence, we analyzed a comprehensive data set of treated and untreated patients included in the registry of the German CLL Study Group (GCLLSG).

The GCLLSG registry is a noninterventional prospective study and all patients with CLL and related malignancies are eligible for participation. Detailed patient, disease, and treatment characteristics are collected in a clinical database by annual visits, SPM are reported on separate forms. All enrolled patients with CLL and an observation time ≥ 24 months from CLL diagnosis were included in this analysis. All malignancies occurring after CLL diagnosis were considered as SPM except for aggressive B-cell lymphomas that were separately reported as Richter's transformations.

Time to SPM was calculated from CLL diagnosis and first-line therapy and evaluated by competing risk analyses considering death as a competing event. Standardized incidence ratios (SIR) from time point of CLL diagnosis were used to compare the SPM incidence in our cohort with that in the German general population with SIR being defined as the ratio of observed SPM and the number of expected SPM based on sex-, age-, and calendar year-matched data from the population-based German cancer registry.

We included 1303 patients with documented CLL diagnosis between 2013 and 2018 in this analysis. At the data cut (June 25, 2020) 480 (36.8%) of these patients had received at least one line of treatment and were categorized as treated while 823 (63.2%) patients were treatment-naïve at last follow-up and allocated to the untreated group. The median observation time from CLL diagnosis was 49.5 months (range, 24.1–88.4) for the full cohort, 55.3 months (24.1–88.1) for the treated cohort and 46.6 months (24.1–88.4) for untreated patients. In the treated group, most patients (78.5%) had only received one line of treatment, 14.8% had two prior lines and 6.7% had three or more CLL-directed therapies. First-line regimens were mostly bendamustine- and fludarabine-based (45.6% and 17.3%), 15.4% had BTK inhibitor-based first-line treatments and 11.7% of all regimens contained chlorambucil (Table S1).

The median age was 68 years (range, 31–90) and 61.1% of the patients were male (Table S2). While pre-existing comorbidities and ECOG performance status were similar between treated and untreated patients, there was a higher proportion of patients with del(17p), unmutated IGHV, and advanced Binet stage in the treated cohort.

Among the 1303 included patients, we observed 105 SPM in 95 patients. In the untreated cohort, 70.4% of all observed SPM were solid malignancies while only 9.3% were hematological malignancies and 20.4% were nonmelanoma skin cancers (NMSC). The distribution in treated patients was 60.8% solid malignancies, 17.6% hematological SPM, and 21.6% NMSC. The most common solid SPM in untreated patients were prostate colorectal and upper gastrointestinal cancer. In the treated cohort, prostate cancer, melanoma, and colorectal cancer occurred most frequently (Table S3).

For competing risk analyses from CLL diagnosis, we considered all 1303 patients untreated until they were censored at the start of first-line treatment. In this cohort, estimated 2-year and 4-year incidences of any SPM were 2.3% and 5.8% (Figure 1A). Treated patients were analyzed separately from the start of first-line treatment and showed 2-year and 4-year incidence rates of any SPM of 5.4% and 9.7% (Figure 1B). To investigate possible associations of specific treatment regimens with the occurrence of SPM, we compared cumulative incidences of different therapies in time-to-SPM analyses. None of the regimens (bendamustine−/fludarabine-/BTK inhibitor−/chlorambucil-based) showed a significantly higher incidence of SPM when compared against all other regimens (Table S4).

We calculated SIR to compare the observed incidence of SPM in our cohort with the incidence in the German general population. The overall risk of SPM in our GCLLSG cohort was not significantly higher than in the matched general population (SIR 1.16, 95% confidence interval [95% CI]: 0.94–1.42) (Table S5); however, a higher-than-expected incidence of hematological SPM was observed in our cohort (SIR 2.18, 95% CI: 1.19–3.65; p = .0053). The incidence of solid SPM and all subcategories of solid malignancies in our GCLLSG cohort showed no difference compared to the general population. While in the untreated cohort, no evaluated malignancy showed a significant increase compared to the general population, the incidence rates of all SPM (SIR 1.42, 95% CI: 1.04–1.90) and in particular hematological SPM (SIR 3.64, 95% CI: 1.66–6.90) were higher than expected in the treated group (Figure 1C).

The few retrospective studies that compared SPM incidences in treated and untreated patients with CLL showed ambiguous results.^{1, 5} While earlier SEER data did not indicate significant differences in these two groups, the latest analysis showed a 38% elevated SPM risk for CLL patients who had received prior chemotherapy compared to a 16% increased risk in those untreated or without available treatment information.¹ Notably, this risk increase was mostly driven by higher numbers of hematological SPM in the treated cohort, similar to our study. The SEER analysis showed a 61% higher risk for hematological malignancies compared to the general population, lower than we had observed in our study (SIR 3.64, 95% CI: 1.66–6.90). However, the SEER analysis assessed a very long time period and describes a significantly higher incidence of hematological SPM in the last studied period (2003–2015) with a SIR of 2.58 (95% CI: 2.32–2.86).¹ The increase of hematological malignancies over time in the SEER data as well as in our data set can likely be attributed to better diagnostic tools but it is also conceivable that the introduction of more intensive treatment options for CLL (e.g. FCR) has contributed to an increase in hematological SPM.^{1, 4} A single-center study from the MDACC identified 797 long-term survivors of CLL, of which 227 were untreated.⁵ The authors compared incidence rates of SPM between the treated and untreated groups and could not detect a significant difference. However, the comparability of our data and the SEER as well as the MDACC data set is impaired by the prospective versus retrospective methodology, different observation periods, and differences in the availability of treatment information.

In line with other studies, we observed a trend toward a higher incidence of melanoma in our treated cohort (SIR 3.53, 95% CI: 0.95–9.04). Increased incidences of melanoma were consistently found in most studies of SPM in CLL and the use of fludarabine-based treatments has been associated with a particularly high risk.^{1, 3}

Our study has various important limitations: By restricting our cohort to patients with exclusively prospective data, the number of patients receiving different treatment regimens is limited, impeding comparisons between these treatment regimens. Furthermore, the above-mentioned differences in the baseline characteristics (genetic risk profile, disease stage) possibly influenced the risk to develop SPM beyond the impact of CLL-directed therapy.

It has been hypothesized that CLL-related immunosuppression is the main driver behind the increased risk for second cancers.^{2, 3, 6} In their study on SPM in BTKi-treated patients, Bond and colleagues demonstrate that higher baseline CD8 counts are associated with a lower risk of developing SPM.³ While we did not obtain data on our patients' cellular and humoral immune status, individuals who developed SPM had a higher incidence of severe infections than those without SPM (p = .032) and more patients in our treated cohort had an advanced disease stage. Both factors suggest a more severe immune dysfunction in the treated cohort that possibly contributed to the higher incidence of SPM.

It has also been discussed that the increase of SPM in patients with CLL is a correlate of more diagnoses of earlier stage cancer due to stringent observation and higher adherence to screening procedures.¹ SPM that are subject to screening programs or tend to be detected by more frequent blood testing or imaging (cancer of prostate, breast, colon, lung) were not increased in our cohort compared to the general population (Figure 1C), thus our study does not support this hypothesis.

In this analysis, we show that the incidence of SPM in treated patients with CLL was 42% higher than in the general population mostly due to a higher-than-expected incidence of hematological SPM while untreated patients with CLL had no increased risk of SPM compared to the general population. No significant differences in SPM incidence could be detected when comparing different first-line regimens. While the underlying mechanisms leading to increased cancer rates in patients with CLL are not yet fully explored, our study suggests that prior CLL-directed treatment has a role in the development of hematological SPM while it does not seem to affect the incidence of most solid malignancies.

随着靶向药物引入慢性淋巴细胞白血病 (CLL) 的治疗范式，患者的治疗效果显着改善。随着以 CLL 患者为主的老年人群的总生存期延长，第二原发恶性肿瘤 (SPM) 作为竞争性死亡原因变得更加相关。先前的分析一致表明，与一般人群相比，在不同治疗背景下，CLL 患者的 SPM 发生率增加。^1-4由于比较治疗和未治疗的 CLL 患者的数据很少，尚不清楚 SPM 的这种增加是否可归因于 CLL 固有的免疫缺陷、CLL 和 SPM 的共同环境或遗传风险因素或抗肿瘤治疗的致癌作用。

为了专门评估 CLL 导向治疗对 SPM 发病率的影响，我们分析了德国 CLL 研究组 (GCLLSG) 登记册中包含的治疗和未治疗患者的综合数据集。

GCLLSG 注册是一项非干预性前瞻性研究，所有 CLL 和相关恶性肿瘤患者都有资格参与。通过年度访问在临床数据库中收集详细的患者、疾病和治疗特征，SPM 以单独的形式报告。所有入选的 CLL 患者和从 CLL 诊断起观察时间≥ 24 个月的患者都包括在该分析中。CLL 诊断后发生的所有恶性肿瘤都被认为是 SPM，除了侵袭性 B 细胞淋巴瘤单独报告为 Richter 转化。

从 CLL 诊断和一线治疗计算到 SPM 的时间，并通过将死亡视为竞争事件的竞争风险分析进行评估。使用 CLL 诊断时间点的标准化发病率 (SIR) 来比较我们队列中的 SPM 发病率与德国一般人群中的 SPM 发生率，其中 SIR 被定义为观察到的 SPM 与基于性别的预期 SPM 数量的比值，来自基于人口的德国癌症登记处的年龄和日历年匹配数据。

我们在该分析中纳入了 2013 年至 2018 年间确诊为 CLL 的 1303 名患者。在数据截止时（2020 年 6 月 25 日），这些患者中有 480 (36.8%) 名接受了至少一种治疗并被归类为已治疗，而 823 (63.2%) 名患者在最后一次随访时未接受过治疗并分配至在未处理团体。从 CLL 诊断开始，整个队列的中位观察时间为 49.5 个月（范围，24.1-88.4），治疗组为 55.3 个月（24.1-88.1），未治疗患者为 46.6 个月（24.1-88.4）。在治疗组中，大多数患者 (78.5%) 只接受过一种治疗，14.8% 有过两种先前的治疗，6.7% 有过三种或更多 CLL 导向疗法。一线方案主要是基于苯达莫司汀和氟达拉滨（45.6% 和 17.3%），15.4% 有基于 BTK 抑制剂的一线治疗，所有方案中有 11.7% 包含苯丁酸氮芥（表 S1）。

中位年龄为 68 岁（范围，31-90），61.1% 的患者为男性（表 S2）。虽然治疗组和未治疗患者的既往合并症和 ECOG 体能状态相似，但在治疗组中，del(17p)、未突变 IGHV 和晚期 Binet 分期的患者比例更高。

在纳入的 1303 名患者中，我们在 95 名患者中观察到 105 个 SPM。在未治疗的队列中，所有观察到的 SPM 中有 70.4% 是实体恶性肿瘤，而只有 9.3% 是血液系统恶性肿瘤，20.4% 是非黑色素瘤皮肤癌 (NMSC)。接受治疗的患者中的分布为 60.8% 的实体恶性肿瘤、17.6% 的血液学 SPM 和 21.6% 的 NMSC。未经治疗的患者中最常见的固体 SPM 是前列腺结直肠癌和上消化道癌。在接受治疗的队列中，前列腺癌、黑色素瘤和结直肠癌的发生频率最高（表 S3）。

对于来自 CLL 诊断的竞争性风险分析，我们考虑了所有 1303 名患者，直到他们在一线治疗开始时被审查。在该队列中，任何 SPM 的估计 2 年和 4 年发生率分别为 2.3% 和 5.8%（图 1A）。接受治疗的患者从一线治疗开始时分别进行分析，结果显示任何 SPM 的 2 年和 4 年发病率分别为 5.4% 和 9.7%（图 1B）。为了研究特定治疗方案与 SPM 发生之间的可能关联，我们比较了不同治疗方法在 SPM 时间分析中的累积发生率。与所有其他方案相比，所有方案（苯达莫司汀-/氟达拉滨-/BTK 抑制剂-/苯丁酸氮芥）均未显示出显着更高的 SPM 发生率（表 S4）。

我们计算了 SIR 以比较我们队列中观察到的 SPM 发生率与德国一般人群的发生率。在我们的 GCLLSG 队列中，SPM 的总体风险并未显着高于匹配的一般人群（SIR 1.16，95% 置信区间 [95% CI]：0.94–1.42）（表 S5）；然而，在我们的队列中观察到血液学 SPM 的发生率高于预期（SIR 2.18，95% CI：1.19-3.65；p = .0053）。与一般人群相比，我们 GCLLSG 队列中实体 SPM 和实体恶性肿瘤的所有亚类的发生率没有显示出差异。虽然在未经治疗的队列中，与一般人群相比，没有评估的恶性肿瘤显示显着增加，所有 SPM（SIR 1.42，95% CI：1.04-1.90）的发生率，特别是血液学 SPM（SIR 3.64，95% CI： 1.66–6.90）高于治疗组的预期（图 1C）。

少数比较治疗和未治疗 CLL 患者 SPM 发生率的回顾性研究显示出模棱两可的结果。^{1, 5}虽然较早的 SEER 数据并未表明这两组之间存在显着差异，但最新分析显示，接受过先前化疗的 CLL 患者的 SPM 风险增加了 38%，而未经治疗或没有可用治疗信息的患者的风险增加了 16% . ¹值得注意的是，与我们的研究类似，这种风险增加主要是由治疗组中较高数量的血液 SPM 驱动的。SEER 分析显示，与一般人群相比，血液系统恶性肿瘤的风险高 61%，低于我们在研究中观察到的风险（SIR 3.64，95% CI：1.66–6.90）。然而，SEER 分析评估了一个很长的时间段，并描述了在最后一个研究时期（2003-2015 年）血液学 SPM 的发生率显着升高，SIR 为 2.58（95% CI：2.32-2.86）。¹SEER 数据以及我们的数据集中血液恶性肿瘤随时间的增加可能归因于更好的诊断工具，但也可以想象，引入更强化的 CLL 治疗方案（例如 FCR）有助于增加在血液学 SPM 中。^{1, 4} MDACC 的一项单中心研究确定了 797 名 CLL 长期幸存者，其中 227 名未经治疗。⁵作者比较了治疗组和未治疗组之间 SPM 的发生率，但未发现显着差异。然而，我们的数据与 SEER 以及 MDACC 数据集的可比性受到前瞻性与回顾性方法、不同观察期以及治疗信息可用性差异的影响。

与其他研究一致，我们观察到治疗队列中黑色素瘤的发病率有更高的趋势（SIR 3.53，95% CI：0.95-9.04）。在 CLL 中的 SPM 的大多数研究中一致发现黑色素瘤的发病率增加，并且使用基于氟达拉滨的治疗与特别高的风险相关。^1、3

我们的研究有各种重要的局限性：通过将我们的队列限制为仅具有前瞻性数据的患者，接受不同治疗方案的患者数量有限，阻碍了这些治疗方案之间的比较。此外，上述基线特征（遗传风险概况、疾病分期）的差异可能会影响发生 SPM 的风险，超出 CLL 导向治疗的影响。

据推测，CLL 相关的免疫抑制是第二种癌症风险增加的主要驱动因素。^{2, 3, 6}在他们对 BTKi 治疗患者的 SPM 的研究中，Bond 及其同事证明，较高的基线 CD8 计数与发生 SPM 的风险较低有关。³虽然我们没有获得关于我们患者的细胞和体液免疫状态的数据，但发生 SPM 的个体比没有 SPM 的个体具有更高的严重感染发生率 ( p  = .032)，并且我们治疗的队列中更多的患者处于晚期疾病阶段. 这两个因素都表明治疗组中更严重的免疫功能障碍可能导致更高的 SPM 发生率。

也有人讨论过，由于严格的观察和对筛查程序的更高依从性，CLL 患者 SPM 的增加与更多早期癌症的诊断相关。¹与一般人群相比，我们的队列中受筛查程序影响或倾向于通过更频繁的血液检测或成像（前列腺癌、乳腺癌、结肠癌、肺癌）检测到的 SPM 没有增加（图 1C），因此我们的研究不支持这个假设。

在该分析中，我们表明，接受治疗的 CLL 患者的 SPM 发生率比一般人群高 42%，主要是由于血液学 SPM 的发生率高于预期，而未接受治疗的 CLL 患者发生 SPM 的风险没有增加对一般人群。比较不同的一线治疗方案时，未检测到 SPM 发生率的显着差异。虽然导致 CLL 患者癌症发病率增加的潜在机制尚未完全探索，但我们的研究表明，先前针对 CLL 的治疗在血液学 SPM 的发展中起作用，而它似乎并不影响大多数实体恶性肿瘤的发病率.